# text-classification

Text classification is the process of assigning predefined categories to text documents using artificial intelligence and machine learning systems.

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## ICARE: cross-domain text classification with incremental class-aware representation and distillation learning

DevFeed: [ICARE: cross-domain text classification with incremental class-aware representation and distillation learning](<https://devfeed.tech/articles/icare-cross-domain-text-classification-with-incremental-class-aware-representation-and-distillation-learning-144992.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.24959>)

Author: Son Duong, Truong X. Tran, Hai-Anh Tran

Published: 2026-11-20T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [text-classification](<https://devfeed.tech/topics/text-classification.md>), [Multitask learning](<https://devfeed.tech/topics/multitask-learning.md>), [Representation learning](<https://devfeed.tech/topics/representation-learning.md>), [Post-training optimization](<https://devfeed.tech/topics/post-training-optimization.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [classification](<https://devfeed.tech/tags/classification.md>), [cross](<https://devfeed.tech/tags/cross.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [datasets](<https://devfeed.tech/tags/datasets.md>), [distillation](<https://devfeed.tech/tags/distillation.md>), [domain](<https://devfeed.tech/tags/domain.md>), [framework](<https://devfeed.tech/tags/framework.md>), [incremental](<https://devfeed.tech/tags/incremental.md>), [inference](<https://devfeed.tech/tags/inference.md>), [knowledge-transfer](<https://devfeed.tech/tags/knowledge-transfer.md>), [language-models](<https://devfeed.tech/tags/language-models.md>), [large-language-models-llms](<https://devfeed.tech/tags/large-language-models-llms.md>), [learning](<https://devfeed.tech/tags/learning.md>), [llms](<https://devfeed.tech/tags/llms.md>), [news](<https://devfeed.tech/tags/news.md>), [overfitting](<https://devfeed.tech/tags/overfitting.md>), [pruning](<https://devfeed.tech/tags/pruning.md>), [representation-learning](<https://devfeed.tech/tags/representation-learning.md>), [sentiment-analysis](<https://devfeed.tech/tags/sentiment-analysis.md>), [text-classification](<https://devfeed.tech/tags/text-classification.md>)

### AI overview

ICARE aims to reduce catastrophic forgetting in continual text classification by combining representation-based instance selection with knowledge distillation across tasks. Its distillation transfers feature-level and logit-level knowledge using pruning techniques and optimized objectives. The abstract reports improved target accuracy and a lower forgetting metric on benchmark text classification datasets, but provides no dataset names, numerical results, or detailed comparison baselines.

### Source excerpt

arXiv:2611.24959v1 Announce Type: cross Abstract: The rise of powerful, fine-tuned large language models (LLMs) necessitates robust natural language inference (NLI) systems. These NLI systems ideally train across various tasks, encompassing diverse concepts like sentiment analysis, news classification, and coreference resolution. However, this continual, multi-tasking approach can lead to catastrophic forgetting, where previously learned tasks are lost, especially in domains lacking specific labels. Rehearsing past examples with labels can further exacerbate this issue by causing overfitting, hindering the system's ability to classify new categories. To tackle these challenges, this research proposes ICARE, a novel framework with two key components: an instance selection module and a knowledge distillation method. The former is based on representation learning, which meticulously selects training examples to present to the NLI system. The latter, knowledge distillation with feature and logit-level transfer, facilitates knowledge transfer between tasks. This leverages pruning techniques and optimized objective functions to transfer both essential feature-level and decision-making (logit-level) knowledge, mitigating catastrophic forgetting while adapting to new patterns. Extensive evaluations on benchmark text classification datasets have demonstrated ICARE's effectiveness. The framework achieves a significant improvement in target accuracy and reduction in the ``forgetting metric''. These results solidify ICARE's ability to preserve and leverage learned knowledge across diverse tasks, ultimately enhancing the adaptability and performance of NLI systems. The source code is available at https://github.com/CongSon01/ICARE_NLP.git

## Don't Pay Attention, PLANT It: Pretraining Attention via Learning-to-Rank

DevFeed: [Don't Pay Attention, PLANT It: Pretraining Attention via Learning-to-Rank](<https://devfeed.tech/articles/don-t-pay-attention-plant-it-pretraining-attention-via-learning-to-rank-143215.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2410.23066>)

Author: Debjyoti Saha Roy, Byron C. Wallace, Javed A. Aslam

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [text-classification](<https://devfeed.tech/topics/text-classification.md>), [Fine-tuning](<https://devfeed.tech/topics/fine-tuning.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [Model Development](<https://devfeed.tech/topics/model-development.md>)

Tags: [ablation](<https://devfeed.tech/tags/ablation.md>), [attention](<https://devfeed.tech/tags/attention.md>), [classification](<https://devfeed.tech/tags/classification.md>), [coding](<https://devfeed.tech/tags/coding.md>), [comparisons](<https://devfeed.tech/tags/comparisons.md>), [content](<https://devfeed.tech/tags/content.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [deepseek](<https://devfeed.tech/tags/deepseek.md>), [deepseek-v3](<https://devfeed.tech/tags/deepseek-v3.md>), [fine-tuning](<https://devfeed.tech/tags/fine-tuning.md>), [learning-to-rank](<https://devfeed.tech/tags/learning-to-rank.md>), [llama-3-1](<https://devfeed.tech/tags/llama-3-1.md>), [llms](<https://devfeed.tech/tags/llms.md>), [medical](<https://devfeed.tech/tags/medical.md>), [mistral](<https://devfeed.tech/tags/mistral.md>), [plug](<https://devfeed.tech/tags/plug.md>), [pretraining](<https://devfeed.tech/tags/pretraining.md>), [recommendation](<https://devfeed.tech/tags/recommendation.md>), [text-classification](<https://devfeed.tech/tags/text-classification.md>), [topic](<https://devfeed.tech/tags/topic.md>), [training](<https://devfeed.tech/tags/training.md>)

### AI overview

PLANT initializes an extreme multi-label classifier's label-attention module by training a Learning-to-Rank model on mutual information gain, then fine-tunes end to end over an LLM's output representations. The plug-and-play approach is reported to outperform state-of-the-art methods on ICD coding, legal topic classification, and content recommendation, and to improve fine-tuned LLMs in controlled nonmedical comparisons. Gains are largest for rare labels, though a training-free LLM extraction baseline remains competitive on rarely seen labels. Ablations identify attention initialization as a key contributor. The supplied text is an abstract, so it does not provide detailed evaluation results.

### Source excerpt

arXiv:2410.23066v5 Announce Type: replace-cross Abstract: State-of-the-art Extreme Multi-Label Text Classification models rely on multi-label attention to focus on key tokens in input text, but learning good attention weights is challenging. We introduce PLANT (Pretrained and Leveraged AtteNTion), a plug-and-play strategy for initializing the label-attention module that operates on top of an LLM's output representations. PLANT works by planting label-specific attention using a Learning-to-Rank model trained on mutual information gain, before end-to-end fine-tuning. This architecture-agnostic approach integrates with LLM backbones (e.g., Mistral-7B, Llama-3.1-8B, DeepSeek-V3, and Phi-3-mini). PLANT outperforms SOTA methods across tasks like ICD coding, legal topic classification, and content recommendation, and improves fine-tuned LLMs in controlled comparisons outside the medical domain. Gains are largest on rare labels, although a training-free LLM extraction baseline remains competitive on rarely seen labels. Ablation studies confirm that attention initialization is a key driver of these gains. We make our code available at https://github.com/deb-official/plant

## Protecting Web Applications using a Multimodal AI-based System for Classifying Web Injection Attacks

DevFeed: [Protecting Web Applications using a Multimodal AI-based System for Classifying Web Injection Attacks](<https://devfeed.tech/articles/protecting-web-applications-using-a-multimodal-ai-based-system-for-classifying-web-injection-attacks-143126.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.23660>)

Author: Rodrigo Branco, Vinicius Cogo, Ib\\'eria Medeiros

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Cryptography and Security](<https://devfeed.tech/sources/arxiv-cryptography-and-security.md>)

Topics: [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [text-classification](<https://devfeed.tech/topics/text-classification.md>), [Web](<https://devfeed.tech/topics/web.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [analysts](<https://devfeed.tech/tags/analysts.md>), [anomaly-detection](<https://devfeed.tech/tags/anomaly-detection.md>), [application-security](<https://devfeed.tech/tags/application-security.md>), [attacks](<https://devfeed.tech/tags/attacks.md>), [cs-cr](<https://devfeed.tech/tags/cs-cr.md>), [detection](<https://devfeed.tech/tags/detection.md>), [encrypted-traffic](<https://devfeed.tech/tags/encrypted-traffic.md>), [incident-response](<https://devfeed.tech/tags/incident-response.md>), [logs](<https://devfeed.tech/tags/logs.md>), [lsh](<https://devfeed.tech/tags/lsh.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [natural-language-processing](<https://devfeed.tech/tags/natural-language-processing.md>), [netflow](<https://devfeed.tech/tags/netflow.md>), [risk-assessment](<https://devfeed.tech/tags/risk-assessment.md>), [web-applications](<https://devfeed.tech/tags/web-applications.md>)

### AI overview

The system classifies web injection attacks without decrypting traffic in-network: it uses NetFlow metadata to flag anomalous flows, analyzes corresponding web server logs after TLS termination with NLP heuristics and supervised machine learning, then checks classifications through a Locality Sensitive Hashing similarity search engine. The authors report evaluation across eight attack classes, with 91% F1 and 95% accuracy, and say the results can help prioritize alerts. The evidence is an abstract and does not provide details such as the evaluation dataset or per-class performance.

### Source excerpt

arXiv:2611.23660v1 Announce Type: new Abstract: Web applications have been essential to accessing and managing online services. Made available with weaknesses, much due to their rapid development and human error, they are exposed to frequent web injection attacks, leaving security analysts overwhelmed with threats and alerts that often lack critical details for proper risk assessment. Detecting these attacks requires inspecting the payload of network traffic packets, which becomes inaccessible when traffic is encrypted via HTTPS, as encryption hides the content of the traffic. This paper introduces a Multimodal Artificial Intelligence (AI) system to protect online services that rely on web application security by detecting and classifying web injection attacks, and provides security analysts with information to support attack prevention and incident response. The system employs unsupervised Machine Learning (ML) to flag potentially anomalous Network Flows (NetFlows). The corresponding data recorded in Web Server Logs (WebLogs) is then analysed using a heuristic framework based on Natural Language Processing (NLP) and supervised ML to detect and classify web injection attacks. Finally, the resulting classifications are scrutinised using a Similarity Search Engine (SSE) based on Locality Sensitive Hashing (LSH) to validate them. The solution handles encrypted traffic by analysing NetFlows and WebLogs, the latter in their natural, unencrypted form, by leveraging NetFlow-level metadata for anomaly detection and analysing web server logs after TLS termination, thereby avoiding any additional in-network traffic decryption. It was evaluated in the classification of eight web injection attack classes. The results demonstrated its ability to accurately identify attacks and their classes, achieving an F_1 score of 91% and an accuracy of 95%, and offering valuable insights for alert prioritisation.

## Who Writes the Codebook? An Eight-Step Framework for Auditing LLM-Generated Category Systems

DevFeed: [Who Writes the Codebook? An Eight-Step Framework for Auditing LLM-Generated Category Systems](<https://devfeed.tech/articles/who-writes-the-codebook-an-eight-step-framework-for-auditing-llm-generated-category-systems-137507.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.19872>)

Author: Thilo Tamme, Anton Hantel, Jakob Scheppers

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [LLM evaluation / benchmarking](<https://devfeed.tech/topics/llm-evaluation-benchmarking.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [text-classification](<https://devfeed.tech/topics/text-classification.md>), [Ground truth / benchmark quality](<https://devfeed.tech/topics/ground-truth-benchmark-quality.md>)

Tags: [auditing](<https://devfeed.tech/tags/auditing.md>), [categories](<https://devfeed.tech/tags/categories.md>), [coding](<https://devfeed.tech/tags/coding.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [cs-cy](<https://devfeed.tech/tags/cs-cy.md>), [framework](<https://devfeed.tech/tags/framework.md>), [grounded](<https://devfeed.tech/tags/grounded.md>), [interviews](<https://devfeed.tech/tags/interviews.md>), [llm](<https://devfeed.tech/tags/llm.md>), [reliability](<https://devfeed.tech/tags/reliability.md>)

### AI overview

The Qualitative Variance Audit Framework tests how LLMs shape qualitative coding by comparing codebook creation, repeated application, and code distributions under controlled inputs. It uses model-owned and shared-reference codebooks to separate variation across the full pipeline from differences applying a common system. In an application to 668 organisational interviews and 13,166 text units, eight models from five vendors produced 48 codebooks and 270 coding runs. Category structures differed markedly: mean within-model reliability ranged from .76 to .92, while between-model agreement on the shared codebook was about .69. The framework reveals model dependencies before interpretation, but measures reliability and representational consequences--not interpretive validity or theory quality.

### Source excerpt

arXiv:2611.19872v1 Announce Type: new Abstract: Large language models (LLMs) increasingly create the categories through which qualitative evidence is interpreted. Existing coding benchmarks often hold these categories fixed. We propose the Qualitative Variance Audit Framework, an eight-step method that compares codebook creation, repeated application and code distributions under controlled inputs. It combines model-owned and shared-reference codebooks to distinguish variation in the overall pipeline from application differences within a common category system. An application to 668 organisational interviews, 13,166 text units and eight models from five vendors yields 48 codebooks and 270 coding runs. Models create markedly different category structures; mean within-model reliability ranges from .76 to .92, whereas between-model agreement on the shared codebook is about .69. The framework complements LLM-assisted grounded theory by exposing model dependencies before substantive interpretation. It assesses reliability and representational consequences, not interpretive validity or the quality of resulting theory.

## Motor Imagery Classification Using Hybrid Cross-Subject Class Mixup and Selective Attention

DevFeed: [Motor Imagery Classification Using Hybrid Cross-Subject Class Mixup and Selective Attention](<https://devfeed.tech/articles/motor-imagery-classification-using-hybrid-cross-subject-class-mixup-and-selective-attention-139102.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.20853>)

Author: Jiwon Kim, Chae-Won Lee, Jong-Seok Lee

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [text-classification](<https://devfeed.tech/topics/text-classification.md>), [data augmentation](<https://devfeed.tech/topics/data-augmentation.md>)

Tags: [attention](<https://devfeed.tech/tags/attention.md>), [bci](<https://devfeed.tech/tags/bci.md>), [brain-computer-interface](<https://devfeed.tech/tags/brain-computer-interface.md>), [classification](<https://devfeed.tech/tags/classification.md>), [cross](<https://devfeed.tech/tags/cross.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [dataset](<https://devfeed.tech/tags/dataset.md>), [eeg](<https://devfeed.tech/tags/eeg.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [functional](<https://devfeed.tech/tags/functional.md>), [fuse](<https://devfeed.tech/tags/fuse.md>), [hybrid](<https://devfeed.tech/tags/hybrid.md>), [imagery](<https://devfeed.tech/tags/imagery.md>), [infrared](<https://devfeed.tech/tags/infrared.md>), [modal](<https://devfeed.tech/tags/modal.md>), [model](<https://devfeed.tech/tags/model.md>), [motor](<https://devfeed.tech/tags/motor.md>), [spectroscopy](<https://devfeed.tech/tags/spectroscopy.md>), [temporal](<https://devfeed.tech/tags/temporal.md>), [token](<https://devfeed.tech/tags/token.md>), [training](<https://devfeed.tech/tags/training.md>)

### AI overview

The framework combines cross-subject class mixup and selective attention to improve hybrid EEG-fNIRS motor imagery classification across subjects. Cross-Subject Class Mixup independently selects EEG and fNIRS donor trials from other training subjects within the anchor class, mixing each modality while preserving its label. Selective Attention applies causal temporal self-attention to EEG and uses a classification token to integrate both modalities, without fNIRS attention updates or direct cross-modal temporal-token attention. In leave-one-subject-out evaluation on a public dataset with 28 subjects, the model reports 76.85% mean balanced accuracy and 76.58% macro-F1, the highest mean performance among five reimplemented baselines. The excerpt also says the authors investigate component contributions through mixing, donor-policy, temporal-policy, cross-modal-path, and ablation comparisons.

### Source excerpt

arXiv:2611.20853v1 Announce Type: new Abstract: Motor imagery (MI) enables a brain-computer interface (BCI) to decode movement intentions without overt movement. Hybrid electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS) classification combines complementary signals, but inter-subject variability and the design of information exchange between modalities affect generalization to new subjects. We propose a framework that combines Cross-Subject Class Mixup (CSCM) with Selective Attention. CSCM aims to construct new training data to make the model robust against inter-subject variability. It independently selects EEG and fNIRS donor trials from other training subjects within the anchor class and mixes each modality while preserving the original label. Selective Attention provides an effective way to simultaneously model the temporal signal characteristics and fuse the two modalities. It retains causal temporal self-attention for EEG and uses a classification token (CLS) to integrate both modalities, without fNIRS attention updates or direct cross-modal temporal-token attention. Leave-one-subject-out (LOSO) evaluation across 28 subjects on a public EEG-fNIRS dataset yields a mean balanced accuracy (BA) of 76.85% and a macro-averaged F1 score (Macro-F1) of 76.58%. The proposed model achieves the highest mean performance among five reimplemented baselines. We also examine the contributions of CSCM and Selective Attention through modality-specific mixing and donor-policy comparisons, temporal-policy comparisons, interventions on direct cross-modal paths, and component ablations.

## Serving Scikit-LLM Pipelines as REST APIs with FastAPI

DevFeed: [Serving Scikit-LLM Pipelines as REST APIs with FastAPI](<https://devfeed.tech/articles/serving-scikit-llm-pipelines-as-rest-apis-with-fastapi-135526.md>)

Original publisher: [Read original article](<https://machinelearningmastery.com/serving-scikit-llm-pipelines-as-rest-apis-with-fastapi/>)

Author: Iván Palomares Carrascosa

Published: 2026-11-16T11:00:00Z

Content type: tutorial

Language: en

Sources: [MachineLearningMastery.com](<https://devfeed.tech/sources/machinelearningmastery-com.md>)

Topics: [FastAPI](<https://devfeed.tech/topics/fastapi.md>), [text-classification](<https://devfeed.tech/topics/text-classification.md>), [REST API](<https://devfeed.tech/topics/rest-api.md>), [Ollama](<https://devfeed.tech/topics/ollama.md>), [scikit-learn](<https://devfeed.tech/topics/scikit-learn.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>)

Tags: [api](<https://devfeed.tech/tags/api.md>), [apis](<https://devfeed.tech/tags/apis.md>), [classification](<https://devfeed.tech/tags/classification.md>), [fastapi](<https://devfeed.tech/tags/fastapi.md>), [language-models](<https://devfeed.tech/tags/language-models.md>), [llm](<https://devfeed.tech/tags/llm.md>), [local](<https://devfeed.tech/tags/local.md>), [ollama](<https://devfeed.tech/tags/ollama.md>), [pipeline](<https://devfeed.tech/tags/pipeline.md>), [pipelines](<https://devfeed.tech/tags/pipelines.md>), [rest](<https://devfeed.tech/tags/rest.md>), [rest-api](<https://devfeed.tech/tags/rest-api.md>), [rest-apis](<https://devfeed.tech/tags/rest-apis.md>), [serve](<https://devfeed.tech/tags/serve.md>), [serving](<https://devfeed.tech/tags/serving.md>), [text](<https://devfeed.tech/tags/text.md>), [text-classification](<https://devfeed.tech/tags/text-classification.md>), [zero-shot](<https://devfeed.tech/tags/zero-shot.md>)

### AI overview

The excerpt describes serving a Scikit-LLM zero-shot text classifier locally through FastAPI and Ollama. It configures a local Ollama model, builds a classifier with scikit-learn's fit/predict pattern, serializes it to a joblib file, then loads it in an asynchronous FastAPI app with a /predict endpoint that can accept multiple texts. The app is run with uvicorn and tested through FastAPI's interactive docs at /docs. The label set supplied during fit defines the only categories the classifier can return; the LLM's weights are not updated, so inputs are mapped into that closed set. The excerpt presents this as a simple one-stage pipeline and notes that Scikit-LLM can support larger pipelines with preprocessing.

### Source excerpt

In this article, you will learn how to serve a Scikit-LLM zero-shot text classification pipeline, powered by a local Ollama model, as a REST API...

## Domain Adaptation Based Pipeline for Character Classification and Handwritten Text Recognition

DevFeed: [Domain Adaptation Based Pipeline for Character Classification and Handwritten Text Recognition](<https://devfeed.tech/articles/domain-adaptation-based-pipeline-for-character-classification-and-handwritten-text-recognition-135846.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.17952>)

Author: Florent Imbert, Simon Corbill\\'e, Hui Han, Elisa H. Barney Smith

Published: 2026-11-16T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [text-classification](<https://devfeed.tech/topics/text-classification.md>), [scikit-learn](<https://devfeed.tech/topics/scikit-learn.md>)

Tags: [classification](<https://devfeed.tech/tags/classification.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [distribution](<https://devfeed.tech/tags/distribution.md>), [domain](<https://devfeed.tech/tags/domain.md>), [pipeline](<https://devfeed.tech/tags/pipeline.md>), [recognition](<https://devfeed.tech/tags/recognition.md>)

### AI overview

The pipeline uses unlabeled target-domain data to adapt a model for character classification and handwritten text recognition when labeled target data is scarce. It first pretrains an expert model on source data, then performs unsupervised domain adaptation, and finally trains the expert on source data in the target style. The authors report that adaptation adds computation during training but not inference. Evaluations cover character classification and historical handwritten-line recognition: the method achieves 76.17% Macro F1 on Safran-MNIST-DLS (DAGECC Track 2), reported as a new state of the art, and competitive results on READ 2018 datasets without target labels.

### Source excerpt

arXiv:2611.17952v1 Announce Type: cross Abstract: Character classification and handwritten text recognition, particularly in specialized tasks, often suffer from limited labeled data, while a large amount of unlabeled data remains unused. To address this challenge, we propose a novel unsupervised domain adaptation-based pipeline to reduce the distribution gap between source and target domains. Our pipeline follows three main steps. In the first step, we pretrained an expert model on source data, to learn a strong general representation. In the second step, we perform unsupervised domain adaptation to leverage unlabeled target data, adapting the model to the specific characteristics of the target domain. In the third step, we trained the expert model on source data in the target style. By aligning the feature distributions between the source and target, our method enables the model to generalize effectively across diverse datasets. The unsupervised training increases the computation time only during training not in inference. We evaluate the pipeline on both character-level and historical handwritten line recognition tasks, demonstrating its flexibility. We establish a new state-of-the-art on the Safran-MNIST-DLS dataset defined in Track 2 of the DAGECC competition with 76.17% Macro F1-score and achieve competitive results on each of the specific datasets of READ 2018 without using the labels of the target data. Code is available at https://github.com/simon-corbi/DRANet-for-classification-and-htr

## Scored Confidence Agent Workflows in the Decision-Making Process

DevFeed: [Scored Confidence Agent Workflows in the Decision-Making Process](<https://devfeed.tech/articles/scored-confidence-agent-workflows-in-the-decision-making-process-134694.md>)

Original publisher: [Read original article](<https://swamp-club.com/feed/p/4e1fbb30-0889-49fb-80a5-0f9c17518481/go>)

Author: cjd

Published: 2026-11-15T18:55:24Z

Content type: article

Language: en

Sources: [Swamp Club Feed](<https://devfeed.tech/sources/swamp-club-feed.md>)

Topics: [text-classification](<https://devfeed.tech/topics/text-classification.md>), [decision-making](<https://devfeed.tech/topics/decision-making.md>), [human review](<https://devfeed.tech/topics/human-review.md>), [agentic workflows](<https://devfeed.tech/topics/agentic-workflows.md>), [AI Integration Strategies](<https://devfeed.tech/topics/ai-integration-strategies.md>), [human in the loop agent](<https://devfeed.tech/topics/human-in-the-loop-agent.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Tool](<https://devfeed.tech/topics/tool.md>), [LLM security](<https://devfeed.tech/topics/llm-security.md>)

Tags: [accountability](<https://devfeed.tech/tags/accountability.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agent-workflows](<https://devfeed.tech/tags/agent-workflows.md>), [agents](<https://devfeed.tech/tags/agents.md>), [classification](<https://devfeed.tech/tags/classification.md>), [confidence](<https://devfeed.tech/tags/confidence.md>), [decision](<https://devfeed.tech/tags/decision.md>), [decision-making](<https://devfeed.tech/tags/decision-making.md>), [executive](<https://devfeed.tech/tags/executive.md>), [humans](<https://devfeed.tech/tags/humans.md>), [in-the-loop](<https://devfeed.tech/tags/in-the-loop.md>), [language](<https://devfeed.tech/tags/language.md>), [language-model](<https://devfeed.tech/tags/language-model.md>), [making](<https://devfeed.tech/tags/making.md>), [model](<https://devfeed.tech/tags/model.md>), [models](<https://devfeed.tech/tags/models.md>), [process](<https://devfeed.tech/tags/process.md>), [swamp](<https://devfeed.tech/tags/swamp.md>), [tasks](<https://devfeed.tech/tags/tasks.md>), [workflows](<https://devfeed.tech/tags/workflows.md>)

### AI overview

The workflow treats suitable language model decisions as classification and leaves high-stakes executive decisions to human review. Deterministic scripts gather accessible facts, while an LLM handles language judgments; its proposed classifications are scored by a separate deterministic system against explicit criteria, then gated before presentation with supporting evidence and criteria met or unmet. Using email sorting as an example, the article lists sender, subject, body, and attachment signals, including numeric and yes-or-no facts. It describes Swamp as the pipeline engine for fetching and parsing data, invoking an agent, structuring its output, applying criteria, and presenting results, with scheduled or triggered runs and built-in encryption for secrets. The author says this approach can use any language model, including Claude, while acknowledging unresolved ethical concerns and the models' unpredictability.

### Source excerpt

A developer describes using scored confidence workflows with any language model via Swamp Club's Swamp framework to handle decision-making as classification, keeping humans in the loop for high-stakes executive decisions while delegating lower-level classification tasks to models.

## When AI Classifies: What Counts as Public Administration?

DevFeed: [When AI Classifies: What Counts as Public Administration?](<https://devfeed.tech/articles/when-ai-classifies-what-counts-as-public-administration-131945.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2607.05420>)

Author: Shaoming Cheng, Laurie A. Schintler

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Databases](<https://devfeed.tech/sources/arxiv-databases.md>)

Topics: [text-classification](<https://devfeed.tech/topics/text-classification.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [human review](<https://devfeed.tech/topics/human-review.md>)

Tags: [administration](<https://devfeed.tech/tags/administration.md>), [ai](<https://devfeed.tech/tags/ai.md>), [classification](<https://devfeed.tech/tags/classification.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cy](<https://devfeed.tech/tags/cs-cy.md>), [cs-db](<https://devfeed.tech/tags/cs-db.md>), [cs-dl](<https://devfeed.tech/tags/cs-dl.md>), [essential](<https://devfeed.tech/tags/essential.md>), [evolution](<https://devfeed.tech/tags/evolution.md>), [identity](<https://devfeed.tech/tags/identity.md>), [knowledge](<https://devfeed.tech/tags/knowledge.md>), [literature](<https://devfeed.tech/tags/literature.md>), [organization](<https://devfeed.tech/tags/organization.md>), [overlap](<https://devfeed.tech/tags/overlap.md>), [patterns](<https://devfeed.tech/tags/patterns.md>), [publication](<https://devfeed.tech/tags/publication.md>), [publishing](<https://devfeed.tech/tags/publishing.md>), [results](<https://devfeed.tech/tags/results.md>), [size](<https://devfeed.tech/tags/size.md>), [structure](<https://devfeed.tech/tags/structure.md>), [study](<https://devfeed.tech/tags/study.md>), [system](<https://devfeed.tech/tags/system.md>)

### AI overview

The study compares five author-defined, citation-driven, and AI-assisted approaches in Web of Science and OpenAlex to examine how they identify public administration and AI-related public administration scholarship. The approaches show limited overlap and differ across corpus size, publication types, DOI availability, outlets, timing, and themes. They tend to represent distinct bodies of work, not simply different-sized subsets of one literature. The authors argue that AI-enabled classification is interpretative and can shape scholarship's visibility and disciplinary boundaries, so expert judgment remains essential and AI assistance should complement it.

### Source excerpt

arXiv:2607.05420v2 Announce Type: replace-cross Abstract: This study examines how alternative systems of scholarly representation identify and characterize public administration (PA) and artificial intelligence-related PA (AI-in-PA) scholarship. The central question is not which system provides a definitive classification of PA, but how alternative systems of scholarly classification differ in identifying and representing its intellectual boundaries. With an exploratory and comparative analysis of Web of Science and OpenAlex, the study compares five approaches based on author-defined, citation-driven, and AI-assisted representations. The results highlight substantial differences and limited overlap in corpus size, publication type, DOI availability, publishing outlets, temporal patterns, and thematic structure. The alternative approaches tend to produce distinct representations because they identify substantially different bodies of scholarship rather than merely larger or smaller subsets of the same literature. The findings suggest that algorithmic knowledge organization influences how interdisciplinary scholarship is classified, connected, and understood and, epistemologically, how its visibility, structure, and intellectual identity are represented. AI-enabled scholarly classifications are not neutral but interpretative, potentially self-reinforcing, and may constrain the evolution and adaptation of disciplinary boundaries. Human expert judgment therefore remains essential and is complemented rather than replaced by AI-assisted scholarly classification.

## Temporal Context-Aware Spam Detection Using Transformer-Based NLP Models

DevFeed: [Temporal Context-Aware Spam Detection Using Transformer-Based NLP Models](<https://devfeed.tech/articles/temporal-context-aware-spam-detection-using-transformer-based-nlp-models-133084.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.16139>)

Author: Fariz Zeynalov, Razim Mammadli

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [text-classification](<https://devfeed.tech/topics/text-classification.md>), [AI Models](<https://devfeed.tech/topics/ai-models.md>), [infini-attention](<https://devfeed.tech/topics/infini-attention.md>)

Tags: [classification](<https://devfeed.tech/tags/classification.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [detection](<https://devfeed.tech/tags/detection.md>), [nlp](<https://devfeed.tech/tags/nlp.md>), [spam](<https://devfeed.tech/tags/spam.md>), [temporal](<https://devfeed.tech/tags/temporal.md>), [transformer](<https://devfeed.tech/tags/transformer.md>)

### AI overview

The abstract reports outbound SMS spam detection at send time using 43,741 messages from a proprietary production database. On an 8,749-message held-out test set, text-only DeBERTa-v3-base outperformed TF-IDF with logistic regression in AUC-ROC (0.9011 vs. 0.8721); DeBERTa's F1 was 0.8412. A deployed DistilBERT model with inter-message temporal features achieved 0.8238 accuracy, 0.8382 spam-class F1, and 0.9128 recall at threshold 0.5 on a separate, class-balanced 20,000-message production evaluation. Those figures are not directly comparable to the held-out results. The paper also describes TemporalBertSpam, which uses a causal Transformer to encode each sender's message history, but says a same-split evaluation is needed to measure its contribution.

### Source excerpt

arXiv:2611.16139v1 Announce Type: cross Abstract: Spam detection is usually treated as single-message classification, ignoring behavioral patterns that emerge across sequences of messages from the same sender. We study outbound SMS spam detection at the carrier level, where the goal is to predict at send time whether a carrier will flag a message as spam. Using a proprietary production database of 43,741 outbound SMS messages with UTC transmission timestamps and carrier delivery statuses, we report three sets of results. On an 8,749-message held-out test set, a TF-IDF + logistic regression baseline reaches AUC-ROC 0.8721, and a text-only DeBERTa-v3-base classifier reaches AUC-ROC 0.9011 and F1 0.8412. A deployed DistilBERT model augmented with inter-message temporal features reaches accuracy 0.8238, spam-class F1 0.8382, and recall 0.9128 at threshold 0.5 on a separate, class-balanced 20,000-message production evaluation set; these results are not directly comparable with the held-out test results. We also describe TemporalBertSpam, an architecture that encodes each sender's message history with a causal Transformer, and outline the same-split evaluation needed to measure its contribution.

## An Extensive Evaluation of Machine Learning and Deep Learning Approaches for Sentiment Analysis: Insights from Amazon Customer Reviews

DevFeed: [An Extensive Evaluation of Machine Learning and Deep Learning Approaches for Sentiment Analysis: Insights from Amazon Customer Reviews](<https://devfeed.tech/articles/an-extensive-evaluation-of-machine-learning-and-deep-learning-approaches-for-sentiment-analysis-insights-from-amazon-customer-reviews-133093.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.16201>)

Author: Muhammed Ihsan Corak, Cuiyuan Wang, Xiaowen Zhang

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [text-classification](<https://devfeed.tech/topics/text-classification.md>), [Deep learning](<https://devfeed.tech/topics/deep-learning.md>), [Embeddings](<https://devfeed.tech/topics/embeddings.md>), [Feature Engineering](<https://devfeed.tech/topics/feature-engineering.md>)

Tags: [amazon](<https://devfeed.tech/tags/amazon.md>), [analysis](<https://devfeed.tech/tags/analysis.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [customer-reviews](<https://devfeed.tech/tags/customer-reviews.md>), [deep-learning](<https://devfeed.tech/tags/deep-learning.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [reviews](<https://devfeed.tech/tags/reviews.md>)

### AI overview

On 568,454 Amazon Fine Food Reviews, Random Forest with Bag-of-Words reached 99% accuracy on a random-oversampled balanced set, while the Attention model reached 98.7%; on the original imbalanced data, Attention achieved 97.8% and FastText 95.3%. The five classical classifiers used Bag-of-Words or TF-IDF capped at 5,000 terms, while six deep learning architectures used learned embeddings over 20,000 words. Deep learning models were reported as more stable across class distributions, while the strongest classical model depended on balancing. Accuracy alone can obscure class-level weaknesses: the RNN scored 91.9% on imbalanced data but had only 0.08 F1 for neutral reviews. Labels came from star ratings, so results reflect agreement with rating-derived classes, and the study uses a single review source.

### Source excerpt

arXiv:2611.16201v1 Announce Type: new Abstract: The rapid growth of user-generated text on online platforms creates opportunities for understanding customer opinion and supporting data-driven decisions. This study compares five traditional machine learning classifiers (Random Forest, Logistic Regression, Bernoulli Naive Bayes, Linear Support Vector Classifier, and Multinomial Naive Bayes) and six deep learning architectures (LSTM, Bi-LSTM, GRU, RNN, Attention, and FastText) for three-class sentiment classification of 568,454 Amazon Fine Food Reviews. Machine learning models used Bag-of-Words (BoW) and TF-IDF features limited to 5,000 terms; deep learning models used learned embeddings over a 20,000-word vocabulary. Models were evaluated on the original imbalanced data and on a random-oversampled balanced set of 1,070,469 reviews, using accuracy, macro and per-class precision, recall, and F1-score, and training time. On the balanced set, Random Forest with BoW achieved 99\% accuracy and the Attention model achieved 98.7\%. Linear SVC reached 89\% accuracy in 1 min 15 s, and Logistic Regression reached 90\% in 3 min 37 s. On the imbalanced set, the Attention model achieved 97.8\% accuracy, and FastText achieved 95.3\% in 3 min 26 s, while the RNN reached 91.9\% with a neutral-class F1-score of 0.08. Deep learning models were more stable across class distributions, whereas the best classical model depended on balancing. We also report a user-facing application that applies the trained Random Forest and Attention models to new review text. Because the labels are derived from star ratings, the results measure agreement with rating-derived sentiment classes rather than annotated sentiment; the dataset is a single source, and the findings should be read with that limitation in mind.

## NLP Models for Text Classification: How to Choose and Implement the Right Approach

DevFeed: [NLP Models for Text Classification: How to Choose and Implement the Right Approach](<https://devfeed.tech/articles/nlp-models-for-text-classification-how-to-choose-and-implement-the-right-approach-131218.md>)

Original publisher: [Read original article](<https://www.toptal.com/developers/machine-learning/nlp-text-classfication-models>)

Author: SHANGLUN WANG, PYTHON DEVELOPER @ TOPTAL

Published: 2026-11-11T05:00:00Z

Content type: article

Language: en

Sources: [Toptal](<https://devfeed.tech/sources/toptal-blog.md>)

Topics: [text-classification](<https://devfeed.tech/topics/text-classification.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [Feature Engineering](<https://devfeed.tech/topics/feature-engineering.md>), [Transformers](<https://devfeed.tech/topics/transformers.md>), [Fine-tuning](<https://devfeed.tech/topics/fine-tuning.md>), [machine learning overfitting](<https://devfeed.tech/topics/machine-learning-overfitting.md>)

Tags: [applications](<https://devfeed.tech/tags/applications.md>), [bert](<https://devfeed.tech/tags/bert.md>), [categories](<https://devfeed.tech/tags/categories.md>), [implement](<https://devfeed.tech/tags/implement.md>), [llms](<https://devfeed.tech/tags/llms.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [models](<https://devfeed.tech/tags/models.md>), [nlp](<https://devfeed.tech/tags/nlp.md>), [resources](<https://devfeed.tech/tags/resources.md>), [text-classification](<https://devfeed.tech/tags/text-classification.md>)

### AI overview

For many text classification tasks, an LLM is a practical starting point: it can assign predefined categories from prompt instructions and recognize varied wording without task-specific feature design. Its fit depends on accuracy and API cost at expected usage. Dedicated models can be preferable when consistent distinctions are difficult for LLMs, when teams need interpretable feature statistics, or at high volume where self-hosting may cost less. Naive Bayes is efficient and works well when individual words signal categories, but handles context poorly; SVMs can be accurate with limited labeled data, though training scales less well and probabilities are not provided by default; logistic regression is inexpensive and interpretable but also struggles with context. Fine-tuned pretrained transformers such as BERT use context and can need less labeled data, but require more compute. The excerpt gives Keras examples for text vectorization and BERT fine-tuning, and recommends precision, recall, and F1 alongside accuracy, especially for imbalanced categories.

### Source excerpt

Text classification can be tackled with a range of approaches, from traditional machine learning techniques to LLMs. Choose an approach that fits the demands of your resources and business case with this guide.

## Counting Under Disagreement: Dollar-Weighted Measurement, Selective Machine Labelling and Budgeted Human Review from Disagreeing AI Readers of Administrative Text

DevFeed: [Counting Under Disagreement: Dollar-Weighted Measurement, Selective Machine Labelling and Budgeted Human Review from Disagreeing AI Readers of Administrative Text](<https://devfeed.tech/articles/counting-under-disagreement-dollar-weighted-measurement-selective-machine-labelling-and-budgeted-human-review-from-disagreeing-ai-readers-of-administrative-text-128271.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.13215>)

Author: George Koomullil (Ascendr, PBC)

Published: 2026-11-11T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [text-classification](<https://devfeed.tech/topics/text-classification.md>), [human review](<https://devfeed.tech/topics/human-review.md>), [Human-AI evaluation](<https://devfeed.tech/topics/human-ai-evaluation.md>), [Simulation](<https://devfeed.tech/topics/simulation.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [budget](<https://devfeed.tech/tags/budget.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [error](<https://devfeed.tech/tags/error.md>), [estimator](<https://devfeed.tech/tags/estimator.md>), [human-review](<https://devfeed.tech/tags/human-review.md>), [independent](<https://devfeed.tech/tags/independent.md>), [language-models](<https://devfeed.tech/tags/language-models.md>), [measurement](<https://devfeed.tech/tags/measurement.md>), [models](<https://devfeed.tech/tags/models.md>), [policy](<https://devfeed.tech/tags/policy.md>), [predictions](<https://devfeed.tech/tags/predictions.md>), [review](<https://devfeed.tech/tags/review.md>), [sample](<https://devfeed.tech/tags/sample.md>), [sampling](<https://devfeed.tech/tags/sampling.md>), [simulation](<https://devfeed.tech/tags/simulation.md>), [stat-me](<https://devfeed.tech/tags/stat-me.md>), [study](<https://devfeed.tech/tags/study.md>), [target](<https://devfeed.tech/tags/target.md>), [tax](<https://devfeed.tech/tags/tax.md>), [text](<https://devfeed.tech/tags/text.md>), [using](<https://devfeed.tech/tags/using.md>)

### AI overview

The study finds that human disagreement materially affects monetary estimates from administrative text: independent readers shift the estimated grant total by $3.93B, while institution-level intervals under-cover at every sample size. Across 121,070 grant-purpose texts and 799,675 records, eight rectified predictors produce totals within $0.4B of one another. For human review, a two-phase policy with a calibrator fitted on its first phase lowers total-estimate RMSE by 28% at a 100-text budget versus dollar-proportional sampling; pure Neyman allocation fails with an uncalibrated predictor. Dollar-weighted machine-label acceptance rules are evaluated, but deployment error control is not established, and a count-based rule accepts no text under dollar-drawn sampling. The findings are rerun under each of three human readers.

### Source excerpt

arXiv:2611.13215v1 Announce Type: new Abstract: We study AI-assisted measurement when one text label applies to many administrative records, the target is a monetary total, and the human readings that anchor the machines disagree. The testbed is 121,070 grant-purpose texts from US tax filings (799,675 records, \$126.4B), labelled by a rule-based classifier and three language models, with a designed sample of 651 texts read independently by three people under one codebook. We state exact conditions for a fixed-predictor difference estimator and for a cross-fitted predictor whose folds are assigned to the whole frame before sampling, under Poisson sampling, and assess the implemented fixed-size procedures by simulation: 8 predictors whose own totals span \$11.2-15.0B agree to within \$0.4B once rectified. Two rules that accept machine labels at a stated dollar-weighted error, weighted conformal p-values under a dollar measure and Learn-then-Test on the design estimate of the error, are evaluated on nested held-out folds and on re-draws of the design with synthetic truths; their error control in deployment is not established, and the count-based rule accepts no text under a dollar-drawn sample. In a replay that charges every human label, a two-phase review policy whose calibrator is fitted on its own first phase reduces the RMSE of the total by 28% at a budget of 100 texts relative to dollar-proportional sampling, while pure Neyman allocation fails with an uncalibrated predictor. Independent readers shift the estimated total by \$3.93B, 2.2 times the within-reader sampling interval, and institution-level intervals under-cover at every size. Every headline result is re-run under each reader. The results identify both the benefits and the limits of using learned predictions to decide what a person should verify.

## Topic-Aware Multi-Task Mental Health Classification with Active Learning-Based Stepwise Relabeling for Social Media Conversations

DevFeed: [Topic-Aware Multi-Task Mental Health Classification with Active Learning-Based Stepwise Relabeling for Social Media Conversations](<https://devfeed.tech/articles/topic-aware-multi-task-mental-health-classification-with-active-learning-based-stepwise-relabeling-for-social-media-conversations-125112.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.09941>)

Author: Revathy Iswarya Sekaran, Vansh Bhardwaj, Aveek Sahana, Utkarshani Jaimini

Published: 2026-11-10T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [text-classification](<https://devfeed.tech/topics/text-classification.md>), [Fine-tuning](<https://devfeed.tech/topics/fine-tuning.md>), [Transformers](<https://devfeed.tech/topics/transformers.md>)

Tags: [classification](<https://devfeed.tech/tags/classification.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [learning](<https://devfeed.tech/tags/learning.md>), [mental-health](<https://devfeed.tech/tags/mental-health.md>), [social-media](<https://devfeed.tech/tags/social-media.md>), [topic](<https://devfeed.tech/tags/topic.md>)

### AI overview

The paper extends a Core Fusion Network to classify mental health condition, emotion, sentiment, and topic together, then uses two-stage active-learning relabeling to refine a small subset of noisy emotion and topic labels. It targets medium-uncertainty samples before high-uncertainty ones and accepts predictions only above a 0.7 confidence threshold to limit error propagation. On 54,412 Reddit posts from five mental health subreddits, the approach with MentalRoBERTa reports gains over the baseline of 0.09% for condition, 0.51% for emotion, 0.35% for sentiment, and 1.57% for topic classification. The abstract attributes added context to topic information; it does not provide further evaluation details.

### Source excerpt

arXiv:2611.09941v1 Announce Type: cross Abstract: Mental health conditions affect millions of people worldwide, yet many cases go undetected until they reach a critical stage. Social media platforms like Reddit have become spaces where people openly share their experiences, but the support is often provided by individuals without clinical training. Existing multi-task approaches jointly model emotion and sentiment for mental health classification but frequently rely on noisy labels generated by pre-trained models and overlook topic information as a useful signal. This paper introduces two key contributions: First, an existing Core Fusion Network framework is extended by adding Topic alongside existing Mental Health Condition, Emotion, and Sentiment Classification tasks, with topic labels derived from the therapist-client interactions in the CounselChat dataset. Second, an Active Learning-based Stepwise Relabeling approach is proposed to refine a small subset of noisy emotion and topic labels using pre-trained models in two stages: targeting medium-uncertainty samples first, followed by high-uncertainty samples. The predictions above a confidence threshold (0.7) are accepted to prevent error propagation. The experiments are conducted on the SuicideWatch and Mental Health dataset of 54,412 Reddit posts collected from five mental health subreddits, evaluating domain-specific transformer encoders including MentalBERT and MentalRoBERTa. The proposed Active Learning-based Stepwise Relabeling, when applied to Core Fusion Network with MentalRoBERTa, improves Mental Health Condition by 0.09%, Emotion by 0.51%, Sentiment by 0.35%, and Topic Classification by 1.57% over the baseline Core Fusion Network. The results suggest that the proposed framework significantly reduces label noise in multi-task mental health classification, and that topic information provides additional context.

## Decision Models in Hugging Face

DevFeed: [Decision Models in Hugging Face](<https://devfeed.tech/articles/decision-models-in-hugging-face-127704.md>)

Original publisher: [Read original article](<https://huggingface.co/blog/decision>)

Author: Aritra Roy Gosthipaty; Lysandre; Pedro Cuenca; Tom Aarsen

Published: 2026-11-10T00:00:00Z

Content type: article

Language: en

Sources: [Hugging Face](<https://devfeed.tech/sources/hugging-face-blog.md>)

Topics: [decision-making](<https://devfeed.tech/topics/decision-making.md>), [Transformers](<https://devfeed.tech/topics/transformers.md>), [text-classification](<https://devfeed.tech/topics/text-classification.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [llama.cpp](<https://devfeed.tech/topics/llama-cpp.md>), [hugging face](<https://devfeed.tech/topics/hugging-face.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [api](<https://devfeed.tech/tags/api.md>), [apis](<https://devfeed.tech/tags/apis.md>), [architectures](<https://devfeed.tech/tags/architectures.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [autoregressive](<https://devfeed.tech/tags/autoregressive.md>), [bernoulli](<https://devfeed.tech/tags/bernoulli.md>), [billing](<https://devfeed.tech/tags/billing.md>), [choice](<https://devfeed.tech/tags/choice.md>), [class](<https://devfeed.tech/tags/class.md>), [classes](<https://devfeed.tech/tags/classes.md>), [classification](<https://devfeed.tech/tags/classification.md>), [classifier](<https://devfeed.tech/tags/classifier.md>), [cost](<https://devfeed.tech/tags/cost.md>), [decision](<https://devfeed.tech/tags/decision.md>), [decision-models](<https://devfeed.tech/tags/decision-models.md>), [hugging-face](<https://devfeed.tech/tags/hugging-face.md>), [jev](<https://devfeed.tech/tags/jev.md>), [models](<https://devfeed.tech/tags/models.md>), [open-source](<https://devfeed.tech/tags/open-source.md>), [science](<https://devfeed.tech/tags/science.md>), [system-one](<https://devfeed.tech/tags/system-one.md>)

### AI overview

Decision models expose structured answers and probabilities directly, avoiding generated response text for tasks with predefined answers. A request pairs a state with typed questions: Choice selects among options, Score returns a probability distribution over ordered levels and its expected index, and Noul estimates the probability of a yes/no statement. The article contrasts this shared interface with fixed-label classifiers, whose classes are built into the head, and zero-shot classifiers, which accept runtime labels but focus on category assignment. It describes support in Transformers through DecisionPipeline, AutoModelForDecision, and DecisionConfig, plus a shared `/v1/systemone` serving API in Transformers and llama.cpp. Compatible autoregressive models can use batched label-logit evaluation without generating text; the article also outlines fine-tuning with Trainer and notes TRL support was still an open PR at publication time.

### Source excerpt

We're on a journey to advance and democratize artificial intelligence through open source and open science.

## Black-box model classification under the discriminative factorization

DevFeed: [Black-box model classification under the discriminative factorization](<https://devfeed.tech/articles/black-box-model-classification-under-the-discriminative-factorization-123135.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2605.07878>)

Author: Hayden Helm, Merrick Ohata, Carey Priebe

Published: 2026-11-09T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [text-classification](<https://devfeed.tech/topics/text-classification.md>), [LLM evaluation / benchmarking](<https://devfeed.tech/topics/llm-evaluation-benchmarking.md>), [Language models](<https://devfeed.tech/topics/language-models.md>)

Tags: [api](<https://devfeed.tech/tags/api.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [auditing](<https://devfeed.tech/tags/auditing.md>), [budget](<https://devfeed.tech/tags/budget.md>), [classification](<https://devfeed.tech/tags/classification.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [embedded](<https://devfeed.tech/tags/embedded.md>), [framework](<https://devfeed.tech/tags/framework.md>), [generative](<https://devfeed.tech/tags/generative.md>), [inference](<https://devfeed.tech/tags/inference.md>), [models](<https://devfeed.tech/tags/models.md>), [oracle](<https://devfeed.tech/tags/oracle.md>), [performance](<https://devfeed.tech/tags/performance.md>), [probability](<https://devfeed.tech/tags/probability.md>), [properties](<https://devfeed.tech/tags/properties.md>), [quality](<https://devfeed.tech/tags/quality.md>), [queries](<https://devfeed.tech/tags/queries.md>), [query](<https://devfeed.tech/tags/query.md>), [querying](<https://devfeed.tech/tags/querying.md>), [responses](<https://devfeed.tech/tags/responses.md>), [set](<https://devfeed.tech/tags/set.md>), [stat-ml](<https://devfeed.tech/tags/stat-ml.md>)

### AI overview

The paper introduces discriminative factorization to assess query-set quality for black-box model classification, where model properties are inferred from embedded responses to API queries. It bounds the probability of chance-level classification by a term that decreases exponentially with the query budget. Across three auditing tasks, estimated factorization parameters predict the observed performance decay rate, and query sets selected using those estimates reproduce the empirical ranking of oracle-selected query sets. The supplied text is an abstract excerpt, so it does not provide details of the tasks or factorization procedure.

### Source excerpt

arXiv:2605.07878v2 Announce Type: replace Abstract: Access to modern generative systems is often restricted to querying an API (the ``black-box'' setting) and many properties of the system are unknown to the user at inference time. Recent work has shown that low-dimensional representations of models based on the relationship of their embedded responses to a set of queries are useful for inferring model-level properties. However, the quality of these representations is highly sensitive to the query set. We introduce the \emph{discriminative factorization} to distinguish between high- and low-quality query sets in the context of black-box model-level classification. Under this framework, the probability of chance-level classification is bounded by a term that decays exponentially in the query budget. On three auditing tasks, estimated factorization parameters predict the empirical performance decay rate. We conclude by showing that query sets selected using the estimated factorization reproduce the empirical ordering of oracle query sets.

## Privacy-Preserving Text Inference for Health-Misinformation Classification and Semantic Retrieval

DevFeed: [Privacy-Preserving Text Inference for Health-Misinformation Classification and Semantic Retrieval](<https://devfeed.tech/articles/privacy-preserving-text-inference-for-health-misinformation-classification-and-semantic-retrieval-119898.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.20796>)

Author: Jasleen Kaur, Chlo\\'e Cachet, Koray Karabina, Jean Belo Klamti, Philippe Lamontagne

Published: 2026-11-06T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Cryptography and Security](<https://devfeed.tech/sources/arxiv-cryptography-and-security.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Cryptography](<https://devfeed.tech/topics/cryptography.md>), [text-classification](<https://devfeed.tech/topics/text-classification.md>), [Embeddings](<https://devfeed.tech/topics/embeddings.md>), [Security & Privacy](<https://devfeed.tech/topics/security-privacy.md>), [information retrieval](<https://devfeed.tech/topics/information-retrieval.md>), [sensitive data](<https://devfeed.tech/topics/sensitive-data.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [analysis](<https://devfeed.tech/tags/analysis.md>), [applications](<https://devfeed.tech/tags/applications.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [based](<https://devfeed.tech/tags/based.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [can](<https://devfeed.tech/tags/can.md>), [challenge](<https://devfeed.tech/tags/challenge.md>), [ckks](<https://devfeed.tech/tags/ckks.md>), [classification](<https://devfeed.tech/tags/classification.md>), [combines](<https://devfeed.tech/tags/combines.md>), [communication](<https://devfeed.tech/tags/communication.md>), [compute](<https://devfeed.tech/tags/compute.md>), [contain](<https://devfeed.tech/tags/contain.md>), [cs-cr](<https://devfeed.tech/tags/cs-cr.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [health](<https://devfeed.tech/tags/health.md>), [inference](<https://devfeed.tech/tags/inference.md>), [privacy](<https://devfeed.tech/tags/privacy.md>), [retrieval](<https://devfeed.tech/tags/retrieval.md>), [semantic](<https://devfeed.tech/tags/semantic.md>), [text](<https://devfeed.tech/tags/text.md>)

### AI overview

The study reports a CKKS homomorphic-encryption framework for classifying health misinformation and retrieving semantically similar text while keeping user queries encrypted during inference on organizational servers. Its classifier fuses BioBERT, PCA-reduced MPNet, and SVD-reduced TF-IDF features, achieving 0.971 accuracy and 0.950 macro-F1 with identical reported plaintext and ciphertext metrics. A retrieval model combines MPNet embeddings and ISCC-SCT fingerprints, correlating 0.8691 with human similarity scores on STS-B. For one query across 7,128 indexed items, SIMD-packed ciphertext retrieval cut latency from 222.028 to 5.168 seconds and communication overhead from 6.11 GB to 2.13 MB versus the naive implementation. The abstract presents these results as evidence of technical feasibility; it does not establish broader deployment performance.

### Source excerpt

arXiv:2610.20796v3 Announce Type: replace Abstract: Health-misinformation analysis presents a dual challenge: misleading claims can affect public well-being, while systems designed to detect them may expose sensitive user-submitted text. This motivates privacy-preserving inference for public-health applications in which user queries may contain confidential medical information and trained models and reference databases are maintained on organizational servers. We design a privacy-preserving text inference framework based on CKKS homomorphic encryption for health-misinformation classification and semantic retrieval. The classification pipeline fuses BioBERT embeddings, PCA-reduced MPNet embeddings, and SVD-reduced TF-IDF features, while the retrieval pipeline combines MPNet embeddings with ISCC-SCT semantic fingerprints through late fusion. We evaluate predictive performance and benchmark plaintext versus ciphertext-based execution. The hybrid classifier achieved an accuracy of 0.971 and a macro-F1 score of 0.950, with identical reported metrics under plaintext and ciphertext-based inference. The hybrid late-fusion similarity model achieved a Pearson correlation of 0.8691 with human sentence-pair similarity scores on STS-B. For ciphertext-based retrieval, we evaluated naive and optimized implementations. The optimized implementation uses CKKS single instruction, multiple data (SIMD) packing to compute and return multiple similarity scores together, reducing single-query latency over 7,128 indexed items from 222.028 to 5.168 seconds and communication overhead from 6.11 GB to 2.13 MB. These findings demonstrate the technical feasibility of privacy-preserving text inference for health-misinformation applications and characterize its predictive utility and efficiency trade-offs.

## HackerSignal: A Multi-Source, CVE-Linked Corpus and Prospective Benchmark for Cyber Threat Intelligence

DevFeed: [HackerSignal: A Multi-Source, CVE-Linked Corpus and Prospective Benchmark for Cyber Threat Intelligence](<https://devfeed.tech/articles/hackersignal-a-multi-source-cve-linked-corpus-and-prospective-benchmark-for-cyber-threat-intelligence-119884.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2605.03158>)

Author: Benjamin M. Ampel, Sagar Samtani

Published: 2026-11-06T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Cryptography and Security](<https://devfeed.tech/sources/arxiv-cryptography-and-security.md>)

Topics: [Threat Hunting & Intel](<https://devfeed.tech/topics/threat-hunting-intel.md>), [OSINT](<https://devfeed.tech/topics/osint.md>), [Vulnerabilities](<https://devfeed.tech/topics/vulnerabilities.md>), [text-classification](<https://devfeed.tech/topics/text-classification.md>)

Tags: [benchmark](<https://devfeed.tech/tags/benchmark.md>), [classification](<https://devfeed.tech/tags/classification.md>), [cs-cr](<https://devfeed.tech/tags/cs-cr.md>), [cve](<https://devfeed.tech/tags/cve.md>), [cyber-threat-intelligence](<https://devfeed.tech/tags/cyber-threat-intelligence.md>), [retrieval](<https://devfeed.tech/tags/retrieval.md>), [temporal](<https://devfeed.tech/tags/temporal.md>)

### AI overview

HackerSignal provides a prospective CTI benchmark for linking documents to vulnerabilities disclosed after training. It aggregates 7.45 million deduplicated documents from 64 public sources across eight layers, keyed to CVEs, and defines temporal-split tasks for CVE retrieval, exploit-type classification, and generalization to non-overlapping CVEs. Zero-shot dense and hybrid retrieval achieved MRR 0.71, with top models statistically tied; cybersecurity-pretrained encoders reached macro-F1 0.90 on exploit classification. An audit found 79.5% precision across 240 linkage labels, with substantial variation by source (20%-100%). Despite 6.97 million hacker-community posts, only about 2,000 were CVE-linked; a forum-grounding pilot achieved MRR 0.08, indicating that linking informal discourse to vulnerabilities remains difficult.

### Source excerpt

arXiv:2605.03158v2 Announce Type: replace Abstract: Cyber threat intelligence (CTI) systems must link exploit code, advisories, and fix commits to the vulnerabilities they concern, and they must do so for vulnerabilities disclosed after training. HackerSignal is a public corpus and benchmark built for this prospective setting. It aggregates 7.45 million exact-deduplicated documents from 64 public sources across eight source layers (hacker forums, exploit archives, exploit question-and-answer sites, vulnerability references, advisories, bug-bounty disclosures, fix commits, and known-exploited-vulnerability records), spanning 1988--2026. All layers are keyed to a shared CVE identifier space with source provenance and release-mode governance. We define three benchmark tasks with temporal splits: (1) CVE linkage retrieval over 340,536 National Vulnerability Database descriptions, (2) exploit type classification into eight classes, and (3) temporal generalization, where training and test CVE sets do not overlap ($C_{\rm train} \cap C_{\rm test} = \emptyset$). Zero-shot dense and hybrid retrievers reach a mean reciprocal rank (MRR) of 0.71 on Task 1, and the top models are statistically tied under paired bootstrap tests. Cybersecurity-pretrained encoders lead exploit type classification (macro-F1 0.90). A two-annotator audit of 240 linkage labels finds 79.5% precision (Cohen's \k{appa} 0.94), with per-source precision from 20% to 100%. The hacker-community layer contributes 6.97 million posts but only about 2,000 CVE-linked records. A masked forum-post grounding pilot over these records reaches MRR 0.08, about nine times lower than the structured-source tasks. Grounding informal hacker discourse to vulnerabilities therefore remains an open problem that the release is designed to support. Code, data, and Croissant metadata are available at https://huggingface.co/datasets/BenAmpel/HackerSignal and https://github.com/BenAmpel/hackersignal.

## Requirements Nobody Wrote Down: Emergence in a Public-Sector ML Pipeline

DevFeed: [Requirements Nobody Wrote Down: Emergence in a Public-Sector ML Pipeline](<https://devfeed.tech/articles/requirements-nobody-wrote-down-emergence-in-a-public-sector-ml-pipeline-120830.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.07951>)

Author: Ronivaldo Ferreira, Vinicius Costa Girard, Cleidson de Souza, Igor Steinmacher, Carla Aguiar, Filipe Roseiro Cogo, Gustavo Pinto

Published: 2026-11-06T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Software Engineering](<https://devfeed.tech/sources/arxiv-software-engineering.md>)

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [text-classification](<https://devfeed.tech/topics/text-classification.md>), [Requirements](<https://devfeed.tech/topics/requirements.md>)

Tags: [acceptance](<https://devfeed.tech/tags/acceptance.md>), [artifacts](<https://devfeed.tech/tags/artifacts.md>), [brazil](<https://devfeed.tech/tags/brazil.md>), [case-study](<https://devfeed.tech/tags/case-study.md>), [categories](<https://devfeed.tech/tags/categories.md>), [classification](<https://devfeed.tech/tags/classification.md>), [cs-se](<https://devfeed.tech/tags/cs-se.md>), [digital](<https://devfeed.tech/tags/digital.md>), [domain](<https://devfeed.tech/tags/domain.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [experimentation](<https://devfeed.tech/tags/experimentation.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [formalized](<https://devfeed.tech/tags/formalized.md>), [goals](<https://devfeed.tech/tags/goals.md>), [government](<https://devfeed.tech/tags/government.md>), [integration](<https://devfeed.tech/tags/integration.md>), [interviews](<https://devfeed.tech/tags/interviews.md>), [ml](<https://devfeed.tech/tags/ml.md>), [pipeline](<https://devfeed.tech/tags/pipeline.md>), [public-sector](<https://devfeed.tech/tags/public-sector.md>), [requirements](<https://devfeed.tech/tags/requirements.md>)

### AI overview

A 15-month case study of a machine-learning classification pipeline for Brazil's digital participation platform found that requirements often emerged during experiments and integration rather than being set upfront. Of 95 reconstructed requirements and criteria, only 40% were formalized; broad institutional goals were endorsed but lacked sponsorship. Government taxonomy determined the classification categories because of its institutional legitimacy. Acceptance combined technical metrics, domain judgment, and political assessment, but lacked shared thresholds and practical validation of usefulness. The authors conclude that technical and institutional artifacts help elicit requirements, and that experimentation and integration are part of requirements engineering for public-sector ML.

### Source excerpt

arXiv:2611.07951v1 Announce Type: new Abstract: Governments increasingly use machine learning (ML) to handle data volumes beyond their analytical capacity, yet little is known about how requirements arise and evolve as an ML experiment moves toward operational use. Requirements Engineering (RE) for ML often assumes that requirements can be elicited upfront; in the public sector, however, demands arrive vague, stakeholders speak different professional languages, and what counts as a ``good'' result is unknown in advance. We report a 15-month case study of a thematic classification pipeline for Brazil's national digital participation platform, which holds over 60,000 citizen proposals, triangulating 43 project artifacts with interviews of six project participants. Institutional demands arrived as broad goals, endorsed but not sponsored, and only 40\% of the 95 reconstructed requirements and criteria were ever formalized. The pipeline's key requirements emerged largely from experiments, model outputs, and integration work, shaped by technical decisions and institutional constraints; the central one, the classification categories, was settled by a government taxonomy chosen for its institutional legitimacy. Acceptance relied on technical metrics, domain judgment, and political assessment, without shared thresholds or validation of usefulness in practice. In public-sector ML, technical and institutional artifacts thus act as elicitation instruments, and experimentation and integration belong to the requirements process.

## Reddit Community Signals as a Supplementary Domain-Reputation Feature in Spam Classification: A Conservative Controlled Evaluation

DevFeed: [Reddit Community Signals as a Supplementary Domain-Reputation Feature in Spam Classification: A Conservative Controlled Evaluation](<https://devfeed.tech/articles/reddit-community-signals-as-a-supplementary-domain-reputation-feature-in-spam-classification-a-conservative-controlled-evaluation-114057.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.01813>)

Author: Fatima Salcedo, Jide Edu

Published: 2026-11-03T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Cryptography and Security](<https://devfeed.tech/sources/arxiv-cryptography-and-security.md>)

Topics: [Detection engineering](<https://devfeed.tech/topics/detection-engineering.md>), [text-classification](<https://devfeed.tech/topics/text-classification.md>), [Feature Engineering](<https://devfeed.tech/topics/feature-engineering.md>), [scikit-learn](<https://devfeed.tech/topics/scikit-learn.md>), [domain](<https://devfeed.tech/topics/domain.md>)

Tags: [classification](<https://devfeed.tech/tags/classification.md>), [community](<https://devfeed.tech/tags/community.md>), [cs-cr](<https://devfeed.tech/tags/cs-cr.md>), [domain](<https://devfeed.tech/tags/domain.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [feature](<https://devfeed.tech/tags/feature.md>), [reddit](<https://devfeed.tech/tags/reddit.md>), [spam](<https://devfeed.tech/tags/spam.md>)

### AI overview

In a preliminary single-split evaluation on 5,582 SpamAssassin-derived emails, Reddit-derived domain signals offered a small, uncertain benefit with Random Forest: Reddit-enhanced features reached 97.94% accuracy, 0.55 percentage points above baseline, but the difference was not statistically significant in the primary test split. The added features were a discussion-presence indicator, keyword-based Reddit score, and urgency interaction. Logistic Regression declined with added features; SVM performed poorly and served as a negative control. Signals appeared most informative for rare domains, yet about 70% of emails had none. Because Reddit data postdated the email corpus, results do not establish real-time detection. The authors call for domain-grouped validation, temporally aligned data, and adversarial robustness testing before production use.

### Source excerpt

arXiv:2611.01813v1 Announce Type: new Abstract: Email spam and phishing remain persistent threats, and purely algorithmic classifiers may struggle with unfamiliar senders, evolving message content, and cold-start domains. This study evaluates whether Reddit-derived community signals can serve as a supplementary domain-reputation feature in email spam classification. We use a preprocessed SpamAssassin-derived corpus of 5,582 emails and compare three feature configurations: (1) a baseline using TF-IDF and structural features; (2) a traditional-enhanced configuration adding urgency and action scores; and (3) a Reddit-enhanced configuration adding a binary discussion-presence indicator, a keyword-based Reddit score, and an urgency-Reddit interaction term. We train Logistic Regression, Random Forest, and Support Vector Machine models. In the preliminary single-split evaluation, the Random Forest model with Reddit-enhanced features achieves 97.94% accuracy, a 0.55 percentage-point improvement over the baseline, although the improvement is not statistically significant in the primary test split. Logistic Regression performance declines when additional features are introduced, and the SVM performs poorly and is treated as a negative control. Reddit signals appear most informative for rare domains, but approximately 70% of emails receive no Reddit-derived signal. Because the Reddit data are temporally later than the email corpus, we do not interpret the results as evidence of real-time detection. Instead, the study provides a conservative evaluation of community-derived signals as a supplementary, asynchronous domain-reputation feature and identifies the additional validation required before production deployment, including domain-grouped validation, temporally aligned contemporary data, and adversarial robustness testing.

## Humanized but Still True? Auditing Facts and Detection in AI-Text Humanizers

DevFeed: [Humanized but Still True? Auditing Facts and Detection in AI-Text Humanizers](<https://devfeed.tech/articles/humanized-but-still-true-auditing-facts-and-detection-in-ai-text-humanizers-114021.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.00359>)

Author: Ziyuan Wang, Genliang Zhu, Xiaobo Chen

Published: 2026-11-03T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Cryptography and Security](<https://devfeed.tech/sources/arxiv-cryptography-and-security.md>)

Topics: [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [text-classification](<https://devfeed.tech/topics/text-classification.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [auditing](<https://devfeed.tech/tags/auditing.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [cs-cr](<https://devfeed.tech/tags/cs-cr.md>), [detection](<https://devfeed.tech/tags/detection.md>), [evasion](<https://devfeed.tech/tags/evasion.md>), [jev](<https://devfeed.tech/tags/jev.md>), [numbers](<https://devfeed.tech/tags/numbers.md>), [retention](<https://devfeed.tech/tags/retention.md>), [rewrite](<https://devfeed.tech/tags/rewrite.md>), [show](<https://devfeed.tech/tags/show.md>), [text](<https://devfeed.tech/tags/text.md>)

### AI overview

The study finds that humanizers often alter checkable details while rewriting AI text. In the October 2026 leaderboard cycle, all 11 commercial tools changed or dropped at least one number, date, name, or quoted span in 35-65% of documents containing such details; most changes involved quotations. An open 12B humanizer retained at least 97% of numbers and names in fact-dense U.S. federal text but reworded 67-78% of quoted spans. Retention did not clearly track detector evasion. For matched federal texts, the zero-shot decision model Jev distinguished human originals from humanized rewrites at AUROC 0.864, outperforming three named detectors by 0.18-0.22, but labeled 24% of human originals AI at the reported threshold. Two perplexity-based detectors performed below chance on AI-styled rewrites. The checker was validated by AI annotators, not human annotation; the supplied material is an abstract.

### Source excerpt

arXiv:2611.00359v1 Announce Type: cross Abstract: AI "humanizers" rewrite machine-generated text so that detectors classify it as human-written. Public leaderboards score evasion and embedding similarity, but not whether checkable details survive. We check whether numbers, dates, person and organization names, and quoted spans survive each rewrite, using a rule-based checker validated by two AI annotators (precision 0.84-1.00, recall 0.93-0.95; no human annotation). In the October 2026 cycle of a public humanizer leaderboard, all 11 commercial tools changed or dropped at least one of these units in 35-65% of the documents that contained one. Most changes were to quoted spans; counting numbers and names alone, 10 of the 11 tools still altered at least one in 20% or more of the documents. The September 2026 cycle was similar (19-50%). An open 12B humanizer kept at least 97% of numbers and names in fact-dense U.S. federal agency text, but reworded 67-78% of the quoted spans, against its own prompt. Unit retention and detector evasion were not clearly related across tools (Spearman rho = -0.46, 95% CI -0.65 to 0.08), and the open model's four released files (two of them quantization-aware low-bit files) did not show the pre-registered monotone trade-off. On federal texts with matched human, AI-styled and humanized versions, a general-purpose decision model, Jev, asked a single zero-shot question, separated human originals from humanized rewrites with AUROC 0.864 (0.827-0.902), 0.18-0.22 higher than Binoculars, Fast-DetectGPT and the adversarially trained RADAR, while calling 24% of the human originals AI at a 0.5 threshold. On the AI-styled rewrites by 2026 models, both perplexity-based detectors scored below chance (AUROC 0.37 and 0.39), also with line breaks removed. Hypotheses, data rules and analyses were frozen before the main runs; two earlier data contacts are disclosed. We release the checker and the per-item results.

## Surrogate Losses for Fairness-Aware Classification: Fisher Consistency and Calibration

DevFeed: [Surrogate Losses for Fairness-Aware Classification: Fisher Consistency and Calibration](<https://devfeed.tech/articles/surrogate-losses-for-fairness-aware-classification-fisher-consistency-and-calibration-114573.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.01613>)

Author: Changcheng Song, Yi Yang, Xiangyu Chang, Xiaoling Lu, Yufeng Liu

Published: 2026-11-03T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [responsible-ai](<https://devfeed.tech/topics/responsible-ai.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [text-classification](<https://devfeed.tech/topics/text-classification.md>)

Tags: [application](<https://devfeed.tech/tags/application.md>), [characterization](<https://devfeed.tech/tags/characterization.md>), [classification](<https://devfeed.tech/tags/classification.md>), [consistency](<https://devfeed.tech/tags/consistency.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [fairness](<https://devfeed.tech/tags/fairness.md>), [loss-functions](<https://devfeed.tech/tags/loss-functions.md>), [objectives](<https://devfeed.tech/tags/objectives.md>), [optimization](<https://devfeed.tech/tags/optimization.md>), [parameter](<https://devfeed.tech/tags/parameter.md>), [stat-ml](<https://devfeed.tech/tags/stat-ml.md>)

### AI overview

The paper studies when surrogate-loss optimization yields decisions that are optimal under fairness-aware risk for demographic parity and equal opportunity, in both attribute-blind and attribute-aware settings. It defines fairness-aware Fisher consistency and characterizes it through derivative conditions, showing that consistency depends on how individual losses interact within the fairness-regularized objective. A calibration inequality connects excess surrogate risk to excess fairness-aware risk. Experiments examine how violations of the consistency conditions affect risk and how the regularization parameter affects whether optimization is well-posed. In a consumer credit default application, the abstract reports that appropriately calibrated fairness regularization substantially reduced demographic disparities while maintaining predictive performance.

### Source excerpt

arXiv:2611.01613v1 Announce Type: cross Abstract: Fairness-aware classification seeks to balance predictive accuracy and fairness across demographic groups. Although surrogate losses facilitate the optimization of fairness-aware objectives, it remains unclear when minimizing these losses yields optimal decisions under the original fairness-aware risk. This paper investigates the statistical validity of surrogate-based fairness-aware classification under demographic parity and equal opportunity, considering both attribute-blind and attribute-aware settings. We introduce fairness-aware Fisher consistency and characterize it through conditions on the derivatives of the surrogate losses. The characterization shows that consistency depends not only on the individual loss functions but also on their interaction within the fairness-regularized objective. We further establish a calibration inequality that relates the excess fairness-aware risk to the excess surrogate risk, providing a quantitative connection between surrogate optimization and the original classification objective. Numerical experiments illustrate how deviations from the consistency conditions affect fairness-aware excess risk and how the regularization parameter influences the well-posedness of surrogate optimization. An application to consumer credit default prediction demonstrates that appropriately calibrated fairness regularization can substantially reduce demographic disparities while maintaining predictive performance.

## Thought-deciphering via hierarchical EEG-to-text alignment

DevFeed: [Thought-deciphering via hierarchical EEG-to-text alignment](<https://devfeed.tech/articles/thought-deciphering-via-hierarchical-eeg-to-text-alignment-114818.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.03623>)

Author: Xiaopeng Mao, Morten M{\\o}rup, Xin Xu, Sang Wan Lee, Sadasivan Puthusserypady

Published: 2026-11-03T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [text-classification](<https://devfeed.tech/topics/text-classification.md>)

Tags: [alignment](<https://devfeed.tech/tags/alignment.md>), [classification](<https://devfeed.tech/tags/classification.md>), [communication](<https://devfeed.tech/tags/communication.md>), [context](<https://devfeed.tech/tags/context.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [decoding](<https://devfeed.tech/tags/decoding.md>), [eeg](<https://devfeed.tech/tags/eeg.md>), [eess-sp](<https://devfeed.tech/tags/eess-sp.md>), [framework](<https://devfeed.tech/tags/framework.md>), [generation](<https://devfeed.tech/tags/generation.md>), [text](<https://devfeed.tech/tags/text.md>)

### AI overview

The paper investigates sentence-level EEG-to-text classification, aiming to retrieve a read sentence from its context and EEG signals using a hierarchical alignment framework. It addresses a gap in research that has focused more on sentence generation and on decoding below the sentence level. The abstract frames EEG as lighter and cheaper than fMRI and MEG, while noting that EEG decoding results remain inferior; it does not provide performance results for the proposed framework.

### Source excerpt

arXiv:2611.03623v1 Announce Type: new Abstract: Communication is important for our well-being, but it can be impaired by disabilities, such as locked-in syndrome (LIS), cerebral palsy (CP), and dysarthria. Non-invasive brain-to-text brain-computer interfaces (BCIs) offer a promising solution by providing an alternative communication pathway for people with communication disabilities. However, most recent breakthroughs have relied on large, cumbersome, and costly neuroimaging modalities, such as functional magnetic resonance imaging (fMRI) and magnetoencephalography (MEG). Numerous studies have attempted decoding using the lighter and cheaper electroencephalography (EEG), but the results remain inferior compared to the other modalities. More importantly, we do not understand how these modalities are related to text at the letter, word, sentence, and higher semantic levels. Many prior studies have focused on decoding below the sentence level because it is considerably easier to design an experiment around phonemes, letters, and words. However, during communication, sentences play a much more important role than words and letters. This is because sentences carry information at such a high level that they can be understood even if some words and letters are missing. However, within sentence-level decoding, sentence classification is not as well investigated as sentence generation, making it difficult to decide whether a sentence should be retrieved or generated in the context of EEG-to-text decoding. This paper strives to investigate sentence-level EEG-to-text classification via a novel hierarchical alignment framework, where the main goal is to retrieve a read sentence using its context and EEG as inputs.

## NormDetect: Anchored Calibration for Low-False-Positive Detection of Machine-Generated Web Text

DevFeed: [NormDetect: Anchored Calibration for Low-False-Positive Detection of Machine-Generated Web Text](<https://devfeed.tech/articles/normdetect-anchored-calibration-for-low-false-positive-detection-of-machine-generated-web-text-111588.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.41459>)

Author: Hanshuai Cui, Fanshuai Meng, Zhiqing Tang

Published: 2026-11-02T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [text-classification](<https://devfeed.tech/topics/text-classification.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [attacks](<https://devfeed.tech/tags/attacks.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [but](<https://devfeed.tech/tags/but.md>), [combines](<https://devfeed.tech/tags/combines.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [detection](<https://devfeed.tech/tags/detection.md>), [false-positive](<https://devfeed.tech/tags/false-positive.md>), [false-positive-rate](<https://devfeed.tech/tags/false-positive-rate.md>), [flag](<https://devfeed.tech/tags/flag.md>), [half](<https://devfeed.tech/tags/half.md>), [language](<https://devfeed.tech/tags/language.md>), [language-models](<https://devfeed.tech/tags/language-models.md>), [large](<https://devfeed.tech/tags/large.md>), [large-language-models](<https://devfeed.tech/tags/large-language-models.md>), [leave](<https://devfeed.tech/tags/leave.md>), [less](<https://devfeed.tech/tags/less.md>), [machine](<https://devfeed.tech/tags/machine.md>), [text](<https://devfeed.tech/tags/text.md>), [web](<https://devfeed.tech/tags/web.md>)

### AI overview

NormDetect combines a normalized zero-shot Binoculars score with a learned per-section head, choosing the blend weight on training folds under a human false-positive cap and calibrating one threshold using human text only. This addresses a limitation of Unicode normalization: it reverses homoglyph edits, but the normalized score still misses many machine-generated texts, including under attacks that leave nothing to clean. On eight RAID benchmark sections across eleven attacks, NormDetect improved recall by 5.0 percentage points over normalized Binoculars at a calibrated 0.1% FPR. Its human FPR bound remained at most 1% in every section, and no other evaluated detector within that bound matched its recall.

### Source excerpt

arXiv:2610.41459v1 Announce Type: new Abstract: Web platforms screen posts for text written by large language models, and every false flag hits a human writer. Detectors must therefore run at a very low false positive rate (FPR). Zero-shot detectors such as Binoculars score text with frozen language models, but invisible edits such as homoglyphs shatter tokenization and push machine text into the human range. Unicode normalization undoes these edits, yet the normalized score still misses half of the machine text at a strict FPR, under attacks that leave nothing to clean. We find that with the Falcon-7B scoring pair, heads on token statistics recall less than the normalized score at this FPR, and several flag too many human posts in some sections. We propose NormDetect, which anchors a learned per-section head to the normalized zero-shot score. A blend weight chosen on training folds under a human FPR cap combines the two scores, and one threshold is calibrated on human text only. On eight sections of the RAID benchmark with eleven attacks, NormDetect raises recall at a 0.1% calibrated FPR by 5.0 points over normalized Binoculars. Its human FPR bound stays at most 1% in every section, and no other detector within this bound recalls as much.

[Next page](<https://devfeed.tech/topics/text-classification.md?cursor=WyIyMDI2LTExLTAyVDA1OjAwOjAwKzAwOjAwIiwgImFhYTUwMmU3LTNjMmQtNGFiOS04Y2MxLTM5MmIxNjMxZWJmZCJd>)