# Data Mining & Modeling

A Google Research Blog label for posts about data mining and modeling.

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

## Simpson's Paradox: Beware the Lurking Variable | Blog | Fivetran

DevFeed: [Simpson's Paradox: Beware the Lurking Variable | Blog | Fivetran](<https://devfeed.tech/articles/simpson-s-paradox-beware-the-lurking-variable-blog-fivetran-140254.md>)

Original publisher: [Read original article](<https://wf.fivetran.com/blog/simpsons-paradox>)

Author: Charles Wang

Published: 2026-11-18T05:58:59Z

Content type: article

Language: en

Sources: [Fivetran Blog](<https://devfeed.tech/sources/fivetran-blog.md>)

Topics: [causality](<https://devfeed.tech/topics/causality.md>), [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>)

Tags: [berkeley](<https://devfeed.tech/tags/berkeley.md>), [blog](<https://devfeed.tech/tags/blog.md>), [business](<https://devfeed.tech/tags/business.md>), [causality](<https://devfeed.tech/tags/causality.md>), [civil-rights](<https://devfeed.tech/tags/civil-rights.md>), [data](<https://devfeed.tech/tags/data.md>), [democratic](<https://devfeed.tech/tags/democratic.md>), [democrats](<https://devfeed.tech/tags/democrats.md>), [department](<https://devfeed.tech/tags/department.md>), [dimensions](<https://devfeed.tech/tags/dimensions.md>), [discrimination](<https://devfeed.tech/tags/discrimination.md>), [earnings](<https://devfeed.tech/tags/earnings.md>), [economy](<https://devfeed.tech/tags/economy.md>), [education](<https://devfeed.tech/tags/education.md>), [election](<https://devfeed.tech/tags/election.md>), [examples](<https://devfeed.tech/tags/examples.md>), [like](<https://devfeed.tech/tags/like.md>), [slice](<https://devfeed.tech/tags/slice.md>), [variable](<https://devfeed.tech/tags/variable.md>), [variables](<https://devfeed.tech/tags/variables.md>)

### AI overview

Simpson's Paradox can make an overall comparison reverse or obscure the patterns within subgroups when group proportions differ. The article illustrates this with Civil Rights Act votes split by region, Berkeley admissions split by department, election preferences linked to race, kidney stone outcomes linked to case severity, and earnings trends affected by changing education levels. For business analysis, it recommends checking relevant dimensions and outside influences, examining data beyond aggregate results, and building a coherent account of how the measurements arose. The examples show why association in aggregated figures alone can mislead; the article does not offer a formal method for establishing causality.

### Source excerpt

You can easily reach misleading conclusions if you don't carefully slice your data, consider lurking variables, and think about causality.

## Solve B2C Attribution Challenges With Fivetran and Braze | Blog | Fivetran

DevFeed: [Solve B2C Attribution Challenges With Fivetran and Braze | Blog | Fivetran](<https://devfeed.tech/articles/solve-b2c-attribution-challenges-with-fivetran-and-braze-blog-fivetran-140257.md>)

Original publisher: [Read original article](<https://wf.fivetran.com/blog/solve-b-2-c-attribution-challenges-with-fivetran-and-braze>)

Author: Fivetran

Published: 2026-11-18T02:46:08Z

Content type: article

Language: en

Sources: [Fivetran Blog](<https://devfeed.tech/sources/fivetran-blog.md>)

Topics: [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>), [data-engineering](<https://devfeed.tech/topics/data-engineering.md>), [Data science and engineering](<https://devfeed.tech/topics/data-science-and-engineering.md>)

Tags: [advertising](<https://devfeed.tech/tags/advertising.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [attribution](<https://devfeed.tech/tags/attribution.md>), [behavior](<https://devfeed.tech/tags/behavior.md>), [braze](<https://devfeed.tech/tags/braze.md>), [campaign](<https://devfeed.tech/tags/campaign.md>), [cloud-data-warehouse](<https://devfeed.tech/tags/cloud-data-warehouse.md>), [connectors](<https://devfeed.tech/tags/connectors.md>), [conversion](<https://devfeed.tech/tags/conversion.md>), [customer-engagement](<https://devfeed.tech/tags/customer-engagement.md>), [data](<https://devfeed.tech/tags/data.md>), [ecommerce](<https://devfeed.tech/tags/ecommerce.md>), [facebook-ads](<https://devfeed.tech/tags/facebook-ads.md>), [fivetran](<https://devfeed.tech/tags/fivetran.md>), [funnel](<https://devfeed.tech/tags/funnel.md>), [google-ads](<https://devfeed.tech/tags/google-ads.md>), [marketing](<https://devfeed.tech/tags/marketing.md>)

### AI overview

The article's approach to B2C attribution is to centralize engagement, advertising and transaction data in a cloud warehouse for analysis. Braze provides detailed customer behavior across devices, channels and touchpoints; Fivetran connectors replicate and normalize that data, with in-warehouse transformations for SQL or BI analysis. The article says ad connectors for services such as Google Ads and Facebook Ads use predefined schemas and recommended configurations, account for restatements and attribution windows, and employ techniques to address sampling error. Payment connectors such as Square, Stripe and Recurly can add conversion data, helping teams connect campaign activity to outcomes and inform ad-spend allocation. The excerpt describes this as an approach, not evidence of measured attribution accuracy or results.

### Source excerpt

In today's marketing environment, B2C attribution is a huge challenge. Fivetran connectors make it easier by effortlessly combining transactional, behavioral and campaign data.

## Automate Your Business's Digital Memory-Building | Blog | Fivetran

DevFeed: [Automate Your Business's Digital Memory-Building | Blog | Fivetran](<https://devfeed.tech/articles/automate-your-business-s-digital-memory-building-blog-fivetran-140236.md>)

Original publisher: [Read original article](<https://wf.fivetran.com/blog/history-mode-feature>)

Author: Fivetran

Published: 2026-11-18T02:45:50Z

Content type: opinion

Language: en

Sources: [Fivetran Blog](<https://devfeed.tech/sources/fivetran-blog.md>)

Topics: [data-engineering](<https://devfeed.tech/topics/data-engineering.md>), [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>), [Database](<https://devfeed.tech/topics/database.md>)

Tags: [2](<https://devfeed.tech/tags/2.md>), [about](<https://devfeed.tech/tags/about.md>), [abstract](<https://devfeed.tech/tags/abstract.md>), [accelerate](<https://devfeed.tech/tags/accelerate.md>), [account](<https://devfeed.tech/tags/account.md>), [aggregation](<https://devfeed.tech/tags/aggregation.md>), [also](<https://devfeed.tech/tags/also.md>), [analysts](<https://devfeed.tech/tags/analysts.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [app](<https://devfeed.tech/tags/app.md>), [applications](<https://devfeed.tech/tags/applications.md>), [automate](<https://devfeed.tech/tags/automate.md>), [basic](<https://devfeed.tech/tags/basic.md>), [better](<https://devfeed.tech/tags/better.md>), [building](<https://devfeed.tech/tags/building.md>), [business](<https://devfeed.tech/tags/business.md>), [digital](<https://devfeed.tech/tags/digital.md>), [fivetran](<https://devfeed.tech/tags/fivetran.md>), [history](<https://devfeed.tech/tags/history.md>), [memory](<https://devfeed.tech/tags/memory.md>)

### AI overview

Fivetran's History Mode preserves changes that current-state tables overwrite: when a value changes, it adds a row for the new values and marks the prior row inactive, turning a type 1 slowly changing dimension into type 2. This lets analysts query how attributes such as account type, content preferences, or customer plan changed over time, including changes that applications may not record for analysis. The author says it can help when more comprehensive instrumentation is not possible and may substitute for instrumentation in simpler cases; the author also acknowledges a favorable bias toward the feature.

### Source excerpt

A Fivetran analyst shares his admittedly biased enthusiasm for History Mode.

## STORMGUIDE-FL: Stochastic Remote Guidance for Efficient Mining of Federated Streaming Data

DevFeed: [STORMGUIDE-FL: Stochastic Remote Guidance for Efficient Mining of Federated Streaming Data](<https://devfeed.tech/articles/stormguide-fl-stochastic-remote-guidance-for-efficient-mining-of-federated-streaming-data-136995.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.20975>)

Author: Manuel R\\"oder, Frank-Michael Schleif

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

Content type: article

Language: en

Sources: [arXiv: Databases](<https://devfeed.tech/sources/arxiv-databases.md>), [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Federated Learning](<https://devfeed.tech/topics/federated-learning.md>), [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>), [Streaming](<https://devfeed.tech/topics/streaming.md>)

Tags: [acquisition](<https://devfeed.tech/tags/acquisition.md>), [asynchronous](<https://devfeed.tech/tags/asynchronous.md>), [bounded](<https://devfeed.tech/tags/bounded.md>), [budget](<https://devfeed.tech/tags/budget.md>), [client](<https://devfeed.tech/tags/client.md>), [clients](<https://devfeed.tech/tags/clients.md>), [compute](<https://devfeed.tech/tags/compute.md>), [concept-drift](<https://devfeed.tech/tags/concept-drift.md>), [continuous](<https://devfeed.tech/tags/continuous.md>), [contributions](<https://devfeed.tech/tags/contributions.md>), [convergence](<https://devfeed.tech/tags/convergence.md>), [cs-db](<https://devfeed.tech/tags/cs-db.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [data-mining](<https://devfeed.tech/tags/data-mining.md>), [data-privacy](<https://devfeed.tech/tags/data-privacy.md>), [devices](<https://devfeed.tech/tags/devices.md>), [federated](<https://devfeed.tech/tags/federated.md>), [guidance](<https://devfeed.tech/tags/guidance.md>), [mining](<https://devfeed.tech/tags/mining.md>), [remote](<https://devfeed.tech/tags/remote.md>), [streaming](<https://devfeed.tech/tags/streaming.md>), [streaming-data](<https://devfeed.tech/tags/streaming-data.md>)

### AI overview

STORMGUIDE-FL targets local concept drift in each client's streaming data and global client drift caused by asynchronous participation and imbalanced contributions. It combines incremental client-side sampling of informative, diverse instances under bounded memory, compute and labeling budgets with server guidance applied as an update-level correction during local optimization. The correction aims to counter client drift without discarding scarce labeled samples. The abstract reports evaluations across gradual, sudden and recurring drift, with sampling gains compounding with the correction and improved speed and robustness of global convergence under resource-constrained, heterogeneous conditions.

### Source excerpt

arXiv:2611.20975v1 Announce Type: cross Abstract: Federated Learning enables collaborative model training across distributed edge devices ("clients") while preserving data privacy. However, real-world deployments involve clients that observe continuous data streams from evolving environments while operating under strict resource limitations. Existing solutions still face two major challenges: local concept drift, where each client's data distribution shifts over time, and global client drift, arising from asynchronous engagement and imbalanced contributions across heterogeneous training participants, formally described as dual client heterogeneity. We propose STORMGUIDE-FL, a federated optimization and data-mining framework that mitigates both types of drift using a server-guided pipeline for data acquisition and local training. For data acquisition, a lightweight client-side sampling strategy incrementally selects informative and diverse instances from non-stationary, non-IID local streams under bounded memory, compute, and labeling budget. For local training, the server's remote guidance is embedded directly into local optimization as an update-level correction, counteracting client drift without discarding scarce labeled samples. We empirically verify the effectiveness of the proposed sampling strategy on federated data streams across gradual, sudden, and recurring concept drift, and demonstrate that its gains compound with the update-level drift correction. Experimental results further show improved speed and robustness of global model convergence under resource-constrained, heterogeneous conditions.

## When Does Oversampling Help? A Decision Framework for Imbalanced Data Streams Under Label Delay and Concept Drift

DevFeed: [When Does Oversampling Help? A Decision Framework for Imbalanced Data Streams Under Label Delay and Concept Drift](<https://devfeed.tech/articles/when-does-oversampling-help-a-decision-framework-for-imbalanced-data-streams-under-label-delay-and-concept-drift-135853.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.17981>)

Author: Ahmed Awad, Anirudh Sengar

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

Content type: article

Language: en

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

Topics: [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>)

Tags: [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [classifiers](<https://devfeed.tech/tags/classifiers.md>), [comparison](<https://devfeed.tech/tags/comparison.md>), [concept-drift](<https://devfeed.tech/tags/concept-drift.md>), [continuous](<https://devfeed.tech/tags/continuous.md>), [cost](<https://devfeed.tech/tags/cost.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [data](<https://devfeed.tech/tags/data.md>), [decision](<https://devfeed.tech/tags/decision.md>), [delay](<https://devfeed.tech/tags/delay.md>), [experimental](<https://devfeed.tech/tags/experimental.md>), [framework](<https://devfeed.tech/tags/framework.md>), [imbalanced-data](<https://devfeed.tech/tags/imbalanced-data.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [predictions](<https://devfeed.tech/tags/predictions.md>), [recommendations](<https://devfeed.tech/tags/recommendations.md>), [recovery](<https://devfeed.tech/tags/recovery.md>), [streaming](<https://devfeed.tech/tags/streaming.md>), [streams](<https://devfeed.tech/tags/streams.md>), [synthetic](<https://devfeed.tech/tags/synthetic.md>), [window](<https://devfeed.tech/tags/window.md>)

### AI overview

Across 15 classifiers tested on six synthetic and six real imbalanced streams, continuous re-evaluation recovered delay-related loss when drift was active, but had no effect on stationary streams. Adding oversampling showed a statistically detectable benefit over re-evaluation in only one real-data configuration. The proposed framework weighs drift density, minority-class learnability, and label delay relative to the refinement window, framing re-evaluation as a recovery channel and oversampling as a minority-coverage channel. Synthetic-stream cost-performance recommendations did not transfer to real streams, whose Pareto fronts varied by drift regime; the authors therefore recommend conditioning method choice on the stream profile and reassessing it as the stream evolves.

### Source excerpt

arXiv:2611.17981v1 Announce Type: new Abstract: For imbalanced data streams with delayed labels, is it worth adding a streaming oversampler, or does continuous re-evaluation of earlier predictions already recover the cost of the delay? We address this through an extensive experimental comparison of streaming oversampling and continuous re-evaluation under joint class imbalance, concept drift, and label delay. Streaming oversamplers are typically evaluated under the test-then-train assumption of immediate labels; their behaviour under these conditions, assessed through re-evaluation, has not been systematically studied. We close this gap with a decision framework that predicts when oversampling adds value over re-evaluation, organised around three axes: drift density, minority-class learnability, and delay relative to the refinement window, and operationalised as a competition between a recovery channel (re-evaluation) and a minority-coverage channel (oversampling). Evaluating $15$ classifiers on $6$ synthetic and $6$ real streams, we find: re-evaluation recovers delay-induced loss on drift-active streams but is silent on stationary ones; oversampling's marginal contribution over re-evaluation is statistically detectable in only one real-data configuration; and cost--performance Pareto recommendations from synthetic streams do not transfer to real data, where fronts partition by drift regime. Method selection should therefore be conditioned on the stream's profile and re-estimated as it evolves, not inherited from synthetic benchmarks.

## Uncertainty-Guided Label Rebalancing for CPS Safety Monitoring

DevFeed: [Uncertainty-Guided Label Rebalancing for CPS Safety Monitoring](<https://devfeed.tech/articles/uncertainty-guided-label-rebalancing-for-cps-safety-monitoring-131667.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2603.25670>)

Author: John Ayotunde, Qinghua Xu, Guancheng Wang, Lionel C. Briand

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

Content type: article

Language: en

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

Topics: [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>), [responsible-ai](<https://devfeed.tech/topics/responsible-ai.md>)

Tags: [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-se](<https://devfeed.tech/tags/cs-se.md>), [cyber-physical-systems](<https://devfeed.tech/tags/cyber-physical-systems.md>), [datasets](<https://devfeed.tech/tags/datasets.md>), [monitoring](<https://devfeed.tech/tags/monitoring.md>), [safety](<https://devfeed.tech/tags/safety.md>), [telemetry](<https://devfeed.tech/tags/telemetry.md>), [uncertainty](<https://devfeed.tech/tags/uncertainty.md>)

### AI overview

U-Balance addresses rare unsafe events in CPS monitoring by using behavioral uncertainty to rebalance training labels before fitting a safety predictor. It trains a GatedMLP on distributional kinematic features from telemetry windows, then probabilistically relabels unusually uncertain safe windows as unsafe, without synthesizing data. On a UAV benchmark, it reports F1/AUROC/AUPRC of 0.801/0.972/0.776; on the quadruped-robot ForEnt dataset, 0.522/0.806/0.323. The abstract reports that it outperforms the strongest safety-prediction baselines on all three metrics in both datasets. With the same Bi-LSTM predictor, uLNR also exceeds the strongest alternative rebalancing methods and Early and Late Fusion on both datasets. Evidence here is limited to the supplied abstract/excerpt.

### Source excerpt

arXiv:2603.25670v5 Announce Type: replace-cross Abstract: Safety monitoring is essential for Cyber-Physical Systems (CPSs) such as Unmanned Aerial Vehicles (UAVs). However, unsafe events are rare in CPS operations, creating class imbalance that degrades data-driven safety predictors. Behavioral uncertainty, reflected in erratic control signals or rapid heading changes, is often correlated with safety outcomes: uncertain behaviors occur more often during unsafe operation. Yet safety monitoring rarely exploits it. We propose U-Balance, a supervised approach that leverages behavioral uncertainty to rebalance datasets before training a safety predictor. U-Balance first trains a GatedMLP-based uncertainty predictor that represents telemetry windows using distributional kinematic features and outputs an uncertainty score. Its uncertainty-guided label rebalancing (uLNR) mechanism probabilistically relabels $\textit{safe}$ windows with unusually high uncertainty as $\textit{unsafe}$, enriching the minority class without synthesizing data. A safety predictor is then trained on the rebalanced dataset. We evaluate U-Balance on a UAV benchmark and ForEnt, a quadruped-robot entrapment dataset. On UAV, U-Balance achieves F1/AUROC/AUPRC of $0.801/0.972/0.776$, outperforming the strongest safety-prediction baselines by 14.3/2.2/6.1 percentage points ($\textit{pp}$), respectively. While the strongest baseline varies across metrics (MLP for F1 and AUPRC; CNN for AUROC), U-Balance leads on all three. On ForEnt, it achieves $0.522/0.806/0.323$, outperforming the strongest baselines by 4.4/9.2/1.4 $\textit{pp}$. Using the same Bi-LSTM safety predictor, uLNR outperforms the strongest alternative rebalancing methods by 17.0/4.1/8.2 $\textit{pp}$ on UAV and 2.3/6.5/1.9 on ForEnt. uLNR also outperforms Early and Late Fusion on both datasets. To our knowledge, this work is the first to use behavioral uncertainty for dataset rebalancing in CPS safety monitoring.

## Google DeepMind Releases AlphaGenome 2 for Cell-Type-Specific Variant Effects

DevFeed: [Google DeepMind Releases AlphaGenome 2 for Cell-Type-Specific Variant Effects](<https://devfeed.tech/articles/google-deepmind-releases-alphagenome-2-for-cell-type-specific-variant-effects-129776.md>)

Original publisher: [Read original article](<https://www.marktechpost.com/2026/11/11/google-deepmind-alphagenome-2-cell-type-variant-effects/>)

Author: Asif Razzaq

Published: 2026-11-11T20:51:56Z

Content type: news

Language: en

Sources: [MarkTechPost](<https://devfeed.tech/sources/marktechpost.md>)

Topics: [Model Development](<https://devfeed.tech/topics/model-development.md>), [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>), [Fine-tuning](<https://devfeed.tech/topics/fine-tuning.md>)

Tags: [agentic-ai](<https://devfeed.tech/tags/agentic-ai.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-paper-summary](<https://devfeed.tech/tags/ai-paper-summary.md>), [ai-shorts](<https://devfeed.tech/tags/ai-shorts.md>), [alphagenome](<https://devfeed.tech/tags/alphagenome.md>), [api](<https://devfeed.tech/tags/api.md>), [applications](<https://devfeed.tech/tags/applications.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [conditional](<https://devfeed.tech/tags/conditional.md>), [deepmind](<https://devfeed.tech/tags/deepmind.md>), [dna](<https://devfeed.tech/tags/dna.md>), [editors-pick](<https://devfeed.tech/tags/editors-pick.md>), [effects](<https://devfeed.tech/tags/effects.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [google](<https://devfeed.tech/tags/google.md>), [google-deepmind](<https://devfeed.tech/tags/google-deepmind.md>), [language-model](<https://devfeed.tech/tags/language-model.md>), [large-language-model](<https://devfeed.tech/tags/large-language-model.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [model](<https://devfeed.tech/tags/model.md>), [new-releases](<https://devfeed.tech/tags/new-releases.md>), [regulatory](<https://devfeed.tech/tags/regulatory.md>), [releases](<https://devfeed.tech/tags/releases.md>), [research](<https://devfeed.tech/tags/research.md>), [sequence](<https://devfeed.tech/tags/sequence.md>), [software-engineering](<https://devfeed.tech/tags/software-engineering.md>), [staff](<https://devfeed.tech/tags/staff.md>), [tech-news](<https://devfeed.tech/tags/tech-news.md>), [technology](<https://devfeed.tech/tags/technology.md>), [type](<https://devfeed.tech/tags/type.md>), [variant](<https://devfeed.tech/tags/variant.md>)

### AI overview

AlphaGenome 2 conditions predictions on one cell state per call, using a 1 Mb human or mouse DNA window. Its condition embedding can be derived from raw single-cell RNA-seq counts, allowing adaptation to unseen datasets without fine-tuning. In the reported benchmarks, it recovered 30% of causal variants at 90% precision on TenK10K single-cell eQTLs, versus 16% for fine-tuned AlphaGenome 1; direction-of-effect recall was 43% versus 23%. The excerpt also reports stronger enhancer-gene prediction than AlphaGenome 1 and Borzoi on held-out CRISPRi data. Access is through a hosted API for non-commercial research; commercial research is in private preview, and model weights are not yet available.

### Source excerpt

Google DeepMind released AlphaGenome 2, a conditional DNA sequence model that predicts regulatory activity and variant effects per cell type. It is available via the AlphaGenome API for non-commercial research. The post Google DeepMind Releases AlphaGenome 2 for Cell-Type-Specific Variant Effects appeared first on MarkTechPost.

## Robust Flow Matching under Target Corruption and Label Noise

DevFeed: [Robust Flow Matching under Target Corruption and Label Noise](<https://devfeed.tech/articles/robust-flow-matching-under-target-corruption-and-label-noise-128691.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.13324>)

Author: Mert Can Kurucu, Erik Englesson, Hossein Azizpour

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

Content type: article

Language: en

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

Topics: [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>), [AI Engineering](<https://devfeed.tech/topics/ai-engineering.md>)

Tags: [analysis](<https://devfeed.tech/tags/analysis.md>), [ccs](<https://devfeed.tech/tags/ccs.md>), [conditional](<https://devfeed.tech/tags/conditional.md>), [corruption](<https://devfeed.tech/tags/corruption.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [field](<https://devfeed.tech/tags/field.md>), [flow](<https://devfeed.tech/tags/flow.md>), [flow-matching](<https://devfeed.tech/tags/flow-matching.md>), [framework](<https://devfeed.tech/tags/framework.md>), [generative](<https://devfeed.tech/tags/generative.md>), [noise](<https://devfeed.tech/tags/noise.md>), [regression](<https://devfeed.tech/tags/regression.md>), [robust](<https://devfeed.tech/tags/robust.md>), [robustness](<https://devfeed.tech/tags/robustness.md>), [score](<https://devfeed.tech/tags/score.md>), [settings](<https://devfeed.tech/tags/settings.md>), [stable](<https://devfeed.tech/tags/stable.md>), [standard](<https://devfeed.tech/tags/standard.md>), [structure](<https://devfeed.tech/tags/structure.md>), [target](<https://devfeed.tech/tags/target.md>)

### AI overview

The paper proposes corruption-specific sample weighting to make Flow Matching more robust to corrupted targets and noisy labels. It models training as time-dependent regression: target corruption adds a residual that grows with time, while label noise directs updates toward the wrong class-conditional vector field. For target corruption, the Temporal Residual Score (TRS) emphasizes residual magnitudes at intermediate-to-late times, where clean and corrupted samples are more separable. For label noise, the Comparative Conditional Score (CCS) compares the observed class field's residual with those of alternative class fields for the same trajectory and velocity target. The abstract reports improved robustness over standard Flow Matching and robust-training baselines in controlled corruption settings and with real-world noisy labels.

### Source excerpt

arXiv:2611.13324v1 Announce Type: new Abstract: Flow Matching (FM) is a strong framework for generative modeling due to its stable training and efficient sampling. However, FM assumes clean target data and labels, an assumption often violated in practice. We formulate FM training as a time-dependent regression problem over conditional vector fields and analyze how the two corruption types affect the objective: i) target corruption adds a time-dependent residual term that grows with time, ii) label noise causes updates to the wrong class-conditional vector field. Based on this analysis, we propose to improve FM robustness by mapping corruption-specific residual scores to sample weights. For target corruption, we propose the Temporal Residual Score (TRS), which emphasizes residual magnitudes at intermediate-to-late FM times, where clean and corrupted samples are more separable. For label noise, we propose the Comparative Conditional Score (CCS), which compares the residual of the observed class-conditional vector field against alternative class-conditional vector fields for the same trajectory and velocity target. Experiments across controlled corruption settings and real-world noisy labels show that the proposed methods improve robustness over standard FM and robust-training baselines, supporting corruption-specific residual structure as a practical basis for robust FM.

## Continuous modeling of manipulative participation for stance analysis: Online discourse on the 2024 Street Animal Law in T\\"urkiye

DevFeed: [Continuous modeling of manipulative participation for stance analysis: Online discourse on the 2024 Street Animal Law in T\\"urkiye](<https://devfeed.tech/articles/continuous-modeling-of-manipulative-participation-for-stance-analysis-online-discourse-on-the-2024-street-animal-law-in-t-urkiye-126142.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.11576>)

Author: Ulya Bayram, Ebru Belk{\\i}s G\\"uzelo\\u{g}lu, Emel Ku\\c{s}ku \\"Ozdemir, Elif \\"Ust\\"unda\\u{g}l{\\i} Erten, \\"Oyk\\"um Esra Yi\\u{g}it

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

Content type: article

Language: en

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

Topics: [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>), [Data analysis](<https://devfeed.tech/topics/data-analysis.md>), [X (Twitter)](<https://devfeed.tech/topics/twitter.md>)

Tags: [case-study](<https://devfeed.tech/tags/case-study.md>), [classification](<https://devfeed.tech/tags/classification.md>), [continuous](<https://devfeed.tech/tags/continuous.md>), [cross-validation](<https://devfeed.tech/tags/cross-validation.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>), [cs-si](<https://devfeed.tech/tags/cs-si.md>), [dataset](<https://devfeed.tech/tags/dataset.md>), [discourse](<https://devfeed.tech/tags/discourse.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [experts](<https://devfeed.tech/tags/experts.md>), [external](<https://devfeed.tech/tags/external.md>), [framework](<https://devfeed.tech/tags/framework.md>), [good](<https://devfeed.tech/tags/good.md>), [gradient](<https://devfeed.tech/tags/gradient.md>), [information](<https://devfeed.tech/tags/information.md>), [integration](<https://devfeed.tech/tags/integration.md>), [law](<https://devfeed.tech/tags/law.md>), [modeling](<https://devfeed.tech/tags/modeling.md>), [online](<https://devfeed.tech/tags/online.md>), [opinion](<https://devfeed.tech/tags/opinion.md>), [political-discourse](<https://devfeed.tech/tags/political-discourse.md>), [protocol](<https://devfeed.tech/tags/protocol.md>)

### AI overview

The study models manipulative participation as a continuous user-level score and combines it with stance analysis, addressing the limits of treating participation as simply authentic or inauthentic. In a case study of online discussion of Türkiye's 2024 Street Animal Law, researchers collected 7.32 million tweets from 537,994 users and had domain experts score 301 selected users with a 10-criterion rubric. Under user-level cross-validation, gradient boosting performed best at estimating manipulation scores, while BERTurk performed best at stance classification; external validation on an independently annotated user set showed good generalization. The abstract supports the approach's usefulness for analyzing the relationship between participation and stance, but does not report detailed metrics.

### Source excerpt

arXiv:2611.11576v1 Announce Type: cross Abstract: Studies of online political discourse commonly infer public opinion through stance analysis while implicitly assuming that observed user participation is authentic. Meanwhile, studies focusing on detecting manipulative participation are typically modeling it as a binary phenomenon, overlooking the varying degrees to which users may engage in coordinated or inauthentic behavior. This study proposes a continuous modeling approach for manipulative participation and examines its integration with stance analysis. Using online discourse on the 2024 Street Animal Law in T\"urkiye as a case study, we collected a dataset comprising 7,320,574 tweets from 537,994 users. From this corpus, 301 users were selected through a high-information sampling strategy and manually annotated by domain experts using a 10-criterion rubric that yields user-level continuous manipulative participation scores. Evaluation under a unified user-level cross-validation protocol showed that the proposed framework accurately estimates both manipulative participation and user stance, with gradient boosting methods providing the best manipulative participation estimates and BERTurk achieving the highest stance classification performance. External validation on an independently annotated user set demonstrated good generalization performance. The findings support modeling manipulative participation as a continuous construct rather than a binary label, providing a more informative basis for computational analyses of the relationship between manipulative participation and stance in online political discourse.

## ShikumiMiner: Mining Recurring Implementation Patterns in AI Codebases

DevFeed: [ShikumiMiner: Mining Recurring Implementation Patterns in AI Codebases](<https://devfeed.tech/articles/shikumiminer-mining-recurring-implementation-patterns-in-ai-codebases-127201.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2609.02789>)

Author: Afsana Tasnim, Saif Uddin Mahmud, Md Erfan, Md Rayhanur Rahman

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

Content type: article

Language: en

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

Topics: [coding patterns](<https://devfeed.tech/topics/coding-patterns.md>), [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>), [Programming Languages / Compilers](<https://devfeed.tech/topics/programming-languages-compilers.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [analysis](<https://devfeed.tech/tags/analysis.md>), [analyze](<https://devfeed.tech/tags/analyze.md>), [c-plus-plus](<https://devfeed.tech/tags/c-plus-plus.md>), [compare](<https://devfeed.tech/tags/compare.md>), [control](<https://devfeed.tech/tags/control.md>), [cs-se](<https://devfeed.tech/tags/cs-se.md>), [framework](<https://devfeed.tech/tags/framework.md>), [functions](<https://devfeed.tech/tags/functions.md>), [github](<https://devfeed.tech/tags/github.md>), [graph](<https://devfeed.tech/tags/graph.md>), [implementation](<https://devfeed.tech/tags/implementation.md>), [language-models](<https://devfeed.tech/tags/language-models.md>), [llm](<https://devfeed.tech/tags/llm.md>), [llm-applications](<https://devfeed.tech/tags/llm-applications.md>), [local-llm](<https://devfeed.tech/tags/local-llm.md>), [mining](<https://devfeed.tech/tags/mining.md>), [open-source](<https://devfeed.tech/tags/open-source.md>), [patterns](<https://devfeed.tech/tags/patterns.md>), [static](<https://devfeed.tech/tags/static.md>), [static-analysis](<https://devfeed.tech/tags/static-analysis.md>)

### AI overview

ShikumiMiner uses static analysis combining Abstract Syntax Tree and Control Flow Graph features to identify and compare recurring implementation patterns in C++ local LLM codebases. The study analyzes ten open-source GitHub repositories and uses a multi-label Random Forest model to classify functions into seven categories defined for the study. The supplied abstract does not specify the categories or report the patterns found.

### Source excerpt

arXiv:2609.02789v2 Announce Type: replace Abstract: Large language models are paving the way towards innovation by understanding, analyzing, summarizing and generating content in the modern world. Currently there are thousands of LLM projects developed by engineers in open-source repositories. However, whether these LLM projects have underlying patterns or not remains a question. Exploring these underlying patterns will give new dimensions to the developers who aim to develop these LLM projects. In this paper, we propose ShikumiMiner, a static-analysis framework that combines Abstract Syntax Tree (AST) and Control Flow Graph (CFG) features to detect and compare recurring implementation patterns in C++ local LLM codebases. We analyze ten GitHub open-source repositories and classify functions into seven study-specific categories using a multi-label Random Forest model. Studying these patterns can provide useful insights for developers aiming to design LLM applications.

## DiSARM: Diverse and Surprising Association Rule Mining with Large Language Models

DevFeed: [DiSARM: Diverse and Surprising Association Rule Mining with Large Language Models](<https://devfeed.tech/articles/disarm-diverse-and-surprising-association-rule-mining-with-large-language-models-125197.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.10445>)

Author: Lijia Liu, Hisashi Kashima

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: [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>), [LLM Techniques](<https://devfeed.tech/topics/llm-techniques.md>)

Tags: [association](<https://devfeed.tech/tags/association.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [data](<https://devfeed.tech/tags/data.md>), [data-analysis](<https://devfeed.tech/tags/data-analysis.md>), [datasets](<https://devfeed.tech/tags/datasets.md>), [diversity](<https://devfeed.tech/tags/diversity.md>), [eda](<https://devfeed.tech/tags/eda.md>), [knowledge](<https://devfeed.tech/tags/knowledge.md>), [language-models](<https://devfeed.tech/tags/language-models.md>), [large-language-models](<https://devfeed.tech/tags/large-language-models.md>), [mining](<https://devfeed.tech/tags/mining.md>), [reliability](<https://devfeed.tech/tags/reliability.md>), [rules](<https://devfeed.tech/tags/rules.md>), [scalability](<https://devfeed.tech/tags/scalability.md>)

### AI overview

DiSARM is a hybrid framework for finding informative association rules in transactional data. It uses large language models' world knowledge to assess subjective interestingness, while drawing on association rule mining to maintain statistical reliability and encourage diversity among selected rules. The abstract says experiments on three transactional datasets show a combination of statistical reliability, surprise, and diversity; it gives no further methodological or comparative details.

### Source excerpt

arXiv:2611.10445v1 Announce Type: cross Abstract: Exploratory Data Analysis (EDA) aims to discover informative patterns in large datasets. For transactional data, Association Rule Mining (ARM) is an established approach that facilitates EDA by efficiently generating candidate patterns. Beyond efficient mining, identifying informative rules has also been extensively studied through both objective and subjective interestingness measures. Recent advances in Large Language Models (LLMs) offer new opportunities for assessing subjective interestingness using world knowledge. Nevertheless, relying solely on LLMs raises concerns about scalability and statistical reliability. Building on prior ARM research, we propose a hybrid framework, DiSARM, that leverages the world knowledge of LLMs to assess the subjective interestingness of rules, while maintaining statistical reliability and promoting diversity among the selected rules. Across three transactional datasets, experiments demonstrate that DiSARM effectively combines statistical reliability with surprise and diversity, offering a promising approach to ADE from transactional data.

## Leakage- and Label-Aware Benchmarking of Class-Imbalance Strategies for Multimodal Chest X-Ray and EHR Classification

DevFeed: [Leakage- and Label-Aware Benchmarking of Class-Imbalance Strategies for Multimodal Chest X-Ray and EHR Classification](<https://devfeed.tech/articles/leakage-and-label-aware-benchmarking-of-class-imbalance-strategies-for-multimodal-chest-x-ray-and-ehr-classification-120684.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.07950>)

Author: Dang-Khoa Bui, Quang-Huy Tran, Minh-Khoi Nguyen-Bui, Duc-Tuan Ngo, Thanh-Trong Tran, Tuan-Khoi Nguyen, Hoang-Anh Ngo

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

Content type: news

Language: en

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

Topics: [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>), [Dataset development](<https://devfeed.tech/topics/dataset-development.md>), [Embeddings](<https://devfeed.tech/topics/embeddings.md>), [scikit-learn](<https://devfeed.tech/topics/scikit-learn.md>)

Tags: [admission](<https://devfeed.tech/tags/admission.md>), [audit](<https://devfeed.tech/tags/audit.md>), [benchmarking](<https://devfeed.tech/tags/benchmarking.md>), [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [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>), [diagnosis](<https://devfeed.tech/tags/diagnosis.md>), [embeddings](<https://devfeed.tech/tags/embeddings.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [multimodal](<https://devfeed.tech/tags/multimodal.md>), [protocol](<https://devfeed.tech/tags/protocol.md>)

### AI overview

A record-level benchmark of class-imbalance strategies for chest X-ray and EHR prediction was misleading: a lookup matching shared EHR vectors nearly matched the best model, while performance on unseen patients fell to the all-positive baseline. The labels help explain this leakage: each patient's labels came from one admission's diagnosis codes and were copied across that patient's radiographs, leaving about a third of images with labels that differed from their own admission codes. With patient-separated evaluation and admission-aligned, rule-refined labels, no tested imbalance strategy improved mean macro F2 over no augmentation by more than 0.01; tuned tree ensembles had no consistent advantage over logistic regression, and released EHR variables remained at baseline. Imaging-time diagnostic history and image embeddings exceeded baseline. Adding either encoder to diagnostic history did not yield a statistically significant F2 gain; exploratory AUROC gains were significant for RAD-DINO, not DenseNet. RAD-DINO had seen 96.6% of the images during self-supervised pretraining, while the CheXpert-trained encoder had not seen them.

### Source excerpt

arXiv:2611.07950v1 Announce Type: new Abstract: Benchmarks of class-imbalance strategies for multimodal chest X-ray (CXR) and electronic health record (EHR) prediction are often evaluated on record-level splits of cohorts in which one patient contributes several radiographs. We audit such a benchmark on 29,083 MIMIC-CXR radiographs of 10,479 patients with eight diagnosis targets. First, the record-split benchmark was misleading: a lookup table that copies the label of a development-set radiograph with the same EHR vector nearly matched the best model, and on unseen patients every model fell to the level of predicting every case positive. Second, tracing the labels back to MIMIC-IV explained why: each patient's labels are keyword matches on the diagnosis codes of a single admission, copied, with one EHR vector per patient, to all of the patient's radiographs, so about a third of the radiographs carry labels that differ from the codes of their own admission. Third, under patient-separated evaluation with admission-aligned and rule-refined labels, the initial conclusions do not hold: in the evaluated configurations no imbalance strategy raised the observed mean macro F2 over no augmentation by more than 0.01, tuned tree ensembles showed no consistent advantage over tuned logistic regression, and the released EHR variables stayed at the level of the all-positive baseline, whereas the diagnostic history available at imaging time and image embeddings exceeded it, including, with rule-refined labels, a CheXpert-trained encoder that never saw these radiographs; RAD-DINO had seen 96.6% of them during self-supervised pretraining. Adding either encoder to the diagnostic history gave no statistically significant F2 gain; exploratory AUROC gains were significant with RAD-DINO but not with DenseNet. We provide reusable diagnostics, a leakage-aware protocol and a checklist.

## Agile Deliberation: How AI helps humans define the gray areas

DevFeed: [Agile Deliberation: How AI helps humans define the gray areas](<https://devfeed.tech/articles/agile-deliberation-how-ai-helps-humans-define-the-gray-areas-118949.md>)

Original publisher: [Read original article](<https://research.google/blog/agile-deliberation-how-ai-helps-humans-define-the-gray-areas/>)

Published: 2026-11-04T18:07:05Z

Content type: article

Language: en

Sources: [The latest research from Google](<https://devfeed.tech/sources/the-latest-research-from-google.md>)

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>), [human computer interaction](<https://devfeed.tech/topics/human-computer-interaction.md>), [ai alignment](<https://devfeed.tech/topics/ai-alignment.md>)

Tags: [1-5](<https://devfeed.tech/tags/1-5.md>), [2026](<https://devfeed.tech/tags/2026.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-alignment](<https://devfeed.tech/tags/ai-alignment.md>), [ai-model](<https://devfeed.tech/tags/ai-model.md>), [algorithms](<https://devfeed.tech/tags/algorithms.md>), [anxiety](<https://devfeed.tech/tags/anxiety.md>), [boundary](<https://devfeed.tech/tags/boundary.md>), [classification](<https://devfeed.tech/tags/classification.md>), [classifiers](<https://devfeed.tech/tags/classifiers.md>), [clustering](<https://devfeed.tech/tags/clustering.md>), [collaboration](<https://devfeed.tech/tags/collaboration.md>), [consistency](<https://devfeed.tech/tags/consistency.md>), [content-moderation](<https://devfeed.tech/tags/content-moderation.md>), [context](<https://devfeed.tech/tags/context.md>), [data-mining](<https://devfeed.tech/tags/data-mining.md>), [data-mining-modeling](<https://devfeed.tech/tags/data-mining-modeling.md>), [datasets](<https://devfeed.tech/tags/datasets.md>), [demo](<https://devfeed.tech/tags/demo.md>), [domain](<https://devfeed.tech/tags/domain.md>), [generative-ai](<https://devfeed.tech/tags/generative-ai.md>), [human-computer-interaction-and-visualization](<https://devfeed.tech/tags/human-computer-interaction-and-visualization.md>), [machine-intelligence](<https://devfeed.tech/tags/machine-intelligence.md>)

### AI overview

Agile Deliberation helps users define subjective image categories through an iterative process: first, they break a broad concept into subcategories and exclusions; then, vision-language models surface borderline images for human feedback. An automatic prompt-optimization loop updates classification boundaries as users clarify their rules. In 18 live sessions lasting 1.5 hours each, participants showed more consistent labeling over time, built classifiers reported to outperform the tested baselines, and reported less fatigue and anxiety than with manual prompt engineering. The authors note that subjective categories lack static ground truth, so offline evaluation cannot capture how a user's definitions evolve; their live study was designed to observe that process. The underlying code and algorithms are open-sourced, and an interactive Colab/Jupyter demo is available, according to the text.

### Source excerpt

Data Mining & Modeling

## Extending the MM-AR Metamodeling Platform with Process Mining Capabilities

DevFeed: [Extending the MM-AR Metamodeling Platform with Process Mining Capabilities](<https://devfeed.tech/articles/extending-the-mm-ar-metamodeling-platform-with-process-mining-capabilities-115967.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.05176>)

Author: Geraldine H\\"urzeler, Hans-Georg Fill, Daniel Borcard

Published: 2026-11-04T05:00:00Z

Content type: article

Language: en

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

Topics: [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>), [3D](<https://devfeed.tech/topics/3d.md>), [Virtual reality](<https://devfeed.tech/topics/virtual-reality.md>), [CSV](<https://devfeed.tech/topics/csv.md>), [Open Source](<https://devfeed.tech/topics/open-source.md>), [Xonsh](<https://devfeed.tech/topics/xonsh.md>)

Tags: [algorithms](<https://devfeed.tech/tags/algorithms.md>), [computer-science](<https://devfeed.tech/tags/computer-science.md>), [cs-db](<https://devfeed.tech/tags/cs-db.md>), [cs-hc](<https://devfeed.tech/tags/cs-hc.md>), [cs-se](<https://devfeed.tech/tags/cs-se.md>), [csv](<https://devfeed.tech/tags/csv.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [event](<https://devfeed.tech/tags/event.md>), [extended-reality](<https://devfeed.tech/tags/extended-reality.md>), [library](<https://devfeed.tech/tags/library.md>), [mining](<https://devfeed.tech/tags/mining.md>), [open-source](<https://devfeed.tech/tags/open-source.md>), [overhead](<https://devfeed.tech/tags/overhead.md>), [process-mining](<https://devfeed.tech/tags/process-mining.md>), [proof-of-concept](<https://devfeed.tech/tags/proof-of-concept.md>), [prototype](<https://devfeed.tech/tags/prototype.md>), [python](<https://devfeed.tech/tags/python.md>), [third-dimension](<https://devfeed.tech/tags/third-dimension.md>), [xr](<https://devfeed.tech/tags/xr.md>)

### AI overview

The proof of concept adds process discovery to MM-AR without changing the platform core: a separate Python service uses PM4Py to create Petri nets and BPMN models from CSV or XES event logs. Across four public logs and 16 configurations, model element counts matched PM4Py's; integration added about 8-53 seconds per run, with overhead growing alongside discovered model size rather than log size. The prototype does not yet use 3D to encode process information.

### Source excerpt

arXiv:2611.05176v1 Announce Type: cross Abstract: Process mining derives process models from event data recorded by information systems, but its results are usually presented as two-dimensional visualizations that are separate from the environments in which enterprise models are created and maintained. Metamodeling platforms for extended reality (XR), such as MM-AR, allow users to define modeling languages and to display models in 2D and 3D, but they do not provide access to event data or mining algorithms so far. In this paper we therefore present a first proof-of-concept for showing how an XR-based metamodeling platform can be extended with process discovery capabilities without changes to the platform core through a separate Python service based on the open source PM4Py process mining library. We demonstrate the approach using three common discovery algorithms for obtaining Petri nets and BPMN models based on CSV or XES event logs. A technical evaluation using four public event logs shows that the element counts of the created model instances match those of the PM4Py models in all 16 configurations. The integration adds between about 8 and 53 seconds per run, and this overhead grows with the size of the discovered model rather than the size of the event log. Although the prototype does not yet use the third dimension to encode process information, it established the foundations for studying such approaches.

## CosmosAlpha: Multi-Relational Graph Search with Pareto-Guided Exploration for LLM-Driven Alpha Mining

DevFeed: [CosmosAlpha: Multi-Relational Graph Search with Pareto-Guided Exploration for LLM-Driven Alpha Mining](<https://devfeed.tech/articles/cosmosalpha-multi-relational-graph-search-with-pareto-guided-exploration-for-llm-driven-alpha-mining-116955.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.04555>)

Author: Zhan-He Wang, Zi-Yang Lang, Yi-Yang Li, Qing-Yu Lan, Yun-Song Piao

Published: 2026-11-04T05:00:00Z

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: [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [Graphs](<https://devfeed.tech/topics/graphs.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [account](<https://devfeed.tech/tags/account.md>), [alpha](<https://devfeed.tech/tags/alpha.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [based](<https://devfeed.tech/tags/based.md>), [best](<https://devfeed.tech/tags/best.md>), [can](<https://devfeed.tech/tags/can.md>), [code](<https://devfeed.tech/tags/code.md>), [correlation](<https://devfeed.tech/tags/correlation.md>), [cross](<https://devfeed.tech/tags/cross.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [csi](<https://devfeed.tech/tags/csi.md>), [data](<https://devfeed.tech/tags/data.md>), [discovery](<https://devfeed.tech/tags/discovery.md>), [diversity](<https://devfeed.tech/tags/diversity.md>), [exploration](<https://devfeed.tech/tags/exploration.md>), [graph](<https://devfeed.tech/tags/graph.md>), [llm](<https://devfeed.tech/tags/llm.md>), [mining](<https://devfeed.tech/tags/mining.md>), [q-fin-cp](<https://devfeed.tech/tags/q-fin-cp.md>), [search](<https://devfeed.tech/tags/search.md>), [with](<https://devfeed.tech/tags/with.md>)

### AI overview

CosmosAlpha searches for predictive financial factors by linking them in a graph that captures generation lineage, expression similarity, and cross-sectional exposure correlation. Its Graph Flow mechanism propagates search utility across related factors, while Pareto-guided selection balances quality, fertility, and diversity. On CSI 300 and CSI 500, CosmosAlpha-Graph achieved the best mean rank across seven predictive and portfolio metrics and outperformed the controlled CosmosAlpha-MCTS counterpart on every reported metric. These results support a consistent advantage for relational graph search in the reported evaluation.

### Source excerpt

arXiv:2611.04555v1 Announce Type: cross Abstract: Discovering predictive alpha factors from noisy financial data is a fundamental yet challenging problem in quantitative finance. LLM-driven formulaic alpha mining offers an attractive paradigm by combining interpretable symbolic expressions with iterative generation and refinement guided by quantitative feedback. However, existing generation-centric and tree-based search strategies primarily organize factors through generation lineage, leaving structural and statistical relationships across search branches underutilized. Consequently, information remains localized along parent--child trajectories, which can confine exploration to redundant regions. Moreover, effective factor discovery must account for multiple distinct objectives that cannot be adequately captured by a single search criterion. We introduce CosmosAlpha, a graph-based LLM factor-mining framework that represents discovered factors as a multi-relational graph encoding generation lineage, expression-structure similarity, and cross-sectional exposure correlation. A Graph Flow mechanism propagates search utilities across the graph, allowing each factor to be assessed using relational evidence from nearby factors rather than in isolation. Pareto-guided selection jointly considers quality, fertility and diversity, explicitly modeling their trade-offs within a multi-objective search framework. Across CSI 300 and CSI 500, CosmosAlpha-Graph achieves the best mean rank over seven predictive and portfolio metrics and outperforms its controlled CosmosAlpha-MCTS counterpart on every reported metric, showing a consistent advantage for relational graph search in LLM-driven factor mining. The source code is available at https://github.com/wzhmtldswz/CosmosAlpha

## An Information Budget for Label-Noise Memorization, and What Statistical Complexity Tracks: A Study with Multilayer Perceptrons

DevFeed: [An Information Budget for Label-Noise Memorization, and What Statistical Complexity Tracks: A Study with Multilayer Perceptrons](<https://devfeed.tech/articles/an-information-budget-for-label-noise-memorization-and-what-statistical-complexity-tracks-a-study-with-multilayer-perceptrons-116914.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.04199>)

Author: Nancy Baygorrea C., Ot\\'avio da Fonseca Martins Gomes

Published: 2026-11-04T05:00:00Z

Content type: article

Language: en

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

Topics: [mlp](<https://devfeed.tech/topics/mlp.md>), [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>), [Occam's razor machine learning](<https://devfeed.tech/topics/occam-s-razor-machine-learning.md>)

Tags: [architecture](<https://devfeed.tech/tags/architecture.md>), [complexity](<https://devfeed.tech/tags/complexity.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [information](<https://devfeed.tech/tags/information.md>), [network](<https://devfeed.tech/tags/network.md>), [noise](<https://devfeed.tech/tags/noise.md>)

### AI overview

Across 89 MNIST runs, normalized memorization information tracked the memorized fraction on training samples, stayed near zero on held-out samples with fresh noisy labels, and was close to a Fano-type bound. Unlike the raw memorized fraction, it has a zero population value under the stated noise-independence condition. Statistical complexity peaked before memorization at corruption rates up to 40%; at higher rates its peak overlapped the memorization phase, near mean training confidence of 0.6. As a stopping rule, it matched established rules only at low noise and detected corrupted samples less well than simple label-free scores. An exploratory 280-run extension found similar behavior under uniform noise, but under pair-flip noise complexity peaked within two epochs. Findings are limited to multilayer perceptrons, two small datasets, and synthetic noise.

### Source excerpt

arXiv:2611.04199v1 Announce Type: new Abstract: How much label noise has a network memorized, and when? We study a multilayer perceptron and what statistical complexity adds to a description of its predictions. Standard identities decompose predictive information into signal, memorization, and heterogeneity. The memorization term vanishes in the population under the stated noise-independence condition. A Fano-type bound relates it to training accuracies, but the memorized fraction M provides no upper bound. Across 89 MNIST runs with ten pre-specified hypotheses (five supported, four unsupported, one not testable), normalized memorization information rose with M on training samples, remained near zero on held-out samples with fresh noisy labels, and was close to its Fano bound. Unlike raw M, it has a zero population null under the independence condition and chance correction largely removed this difference. When every label was replaced by an incorrect class, test accuracy fell to 0.5%, yet predictions retained information about the true class. Mean statistical complexity peaked before memorization at corruption rates up to 40%. At higher rates, its peak fell within the memorization phase, near the epoch when mean training confidence reached 0.6 and the two could not be separated. As a stopping rule, it matched established rules only at low noise and detected corrupted samples less well than simple label-free scores. An exploratory extension of 280 runs covering Fashion-MNIST, pair-flip noise, and four training variants yielded similar results under uniform noise, but not pair-flip noise, where complexity peaked within two epochs. Conclusions are limited to this architecture, two small datasets, and synthetic noise.

## Does reflective LLM feedback survive a wet-lab-scale label budget?

DevFeed: [Does reflective LLM feedback survive a wet-lab-scale label budget?](<https://devfeed.tech/articles/does-reflective-llm-feedback-survive-a-wet-lab-scale-label-budget-115231.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.01908>)

Author: Haoxuan Zeng, Xin Luo

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

Content type: article

Language: en

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

Topics: [Large Language Model](<https://devfeed.tech/topics/llm.md>), [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>)

Tags: [acquisition](<https://devfeed.tech/tags/acquisition.md>), [budget](<https://devfeed.tech/tags/budget.md>), [correlation](<https://devfeed.tech/tags/correlation.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [feedback](<https://devfeed.tech/tags/feedback.md>), [lab](<https://devfeed.tech/tags/lab.md>), [llm](<https://devfeed.tech/tags/llm.md>), [model](<https://devfeed.tech/tags/model.md>), [random](<https://devfeed.tech/tags/random.md>), [rank](<https://devfeed.tech/tags/rank.md>), [recovery](<https://devfeed.tech/tags/recovery.md>)

### AI overview

Across 36 ProteinGym assays with 96 initial labels and six rounds of 96-label acquisition, full LLM control showed no detected test-rank-correlation advantage over random edits (Δz = −0.008; 95% CI −0.034 to +0.018), and equivalence was not established. Exploratory ablations found gains from LLM-proposed model edits: Fisher-transformed test Spearman rose by 0.043 with random acquisition, and normalized top-variant recovery rose by 0.023 with upper-confidence-bound acquisition. LLM acquisition alone did not detectably beat fixed adaptive rules, and repeated feedback did not detectably beat one initial recommendation. The results support constrained model editing in these experiments; added value from recursive reflection at this label budget remains unestablished.

### Source excerpt

arXiv:2611.01908v1 Announce Type: new Abstract: Can reflective language-model feedback improve a biological predictor when each cycle buys only one plate of labels? We separate three possible sources of benefit: proposing model edits, choosing measurements, and revising decisions through repeated feedback. We replay $36$ ProteinGym assays, starting with $96$ labels and acquiring $96$ per round for six rounds, with a sealed test set and a shared edit schema. The pre-specified comparison of full LLM control against random edits detects no advantage in test rank correlation ($\Delta z=-0.008$, $95\%$ CI $[-0.034,+0.018]$); equivalence is also unestablished. Exploratory ablations nevertheless identify useful model editing. With random acquisition fixed, LLM edits improve Fisher-transformed test Spearman by $0.043$ over no edits; with acquisition fixed to upper confidence bound, they improve normalized top-variant recovery by $0.023$. LLM acquisition alone does not detectably outperform fixed adaptive rules, and repeated feedback does not detectably outperform one initial recommendation under full control. Thus, the observed gains support constrained specialist editing, while the incremental value of recursive reflection remains unestablished at this label budget.

## Abstention Spends Its Budget on Local Label Variance More Than on Novelty: Auditing Conformal Calibration and Selective Prediction under Molecular Scaffold Shift

DevFeed: [Abstention Spends Its Budget on Local Label Variance More Than on Novelty: Auditing Conformal Calibration and Selective Prediction under Molecular Scaffold Shift](<https://devfeed.tech/articles/abstention-spends-its-budget-on-local-label-variance-more-than-on-novelty-auditing-conformal-calibration-and-selective-prediction-under-molecular-scaffold-shift-114558.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.01512>)

Author: Arnav Srivastava, Krishna Kassetty

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

Content type: article

Language: en

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

Topics: [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>)

Tags: [auditing](<https://devfeed.tech/tags/auditing.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [prediction](<https://devfeed.tech/tags/prediction.md>), [provenance](<https://devfeed.tech/tags/provenance.md>), [selection](<https://devfeed.tech/tags/selection.md>), [shift](<https://devfeed.tech/tags/shift.md>), [stat-ml](<https://devfeed.tech/tags/stat-ml.md>), [variance](<https://devfeed.tech/tags/variance.md>)

### AI overview

Across a gradient-boosted fingerprint model evaluated under a Bemis-Murcko scaffold split, conformal coverage for lipophilicity depended on calibration provenance: calibration on held-out training-scaffold compounds gave 0.841 coverage against a 0.90 target, while novel-scaffold calibration gave 0.904. BBBP coverage remained above target in all regimes. Abstention scores tracked local label variance more than chemical novelty; for split conformalized quantile regression, its constant correction did not change which compounds were abstained on. At fixed assay budgets, interval-width or ensemble-variance gating did not improve hit precision over point-prediction ranking. A plain novelty gate showed small, uncorrected gains at the smallest budgets. Exploratory exact conformal selection exceeded its false discovery rate target with training-scaffold calibration, but was consistent with the target under novel-scaffold calibration. The reported benefit is calibration-dependent guarantees and batch sizing, rather than improved ranking.

### Source excerpt

arXiv:2611.01512v1 Announce Type: new Abstract: Selective prediction with split conformal calibration promises a post-hoc fix for molecular property prediction: abstain on the least reliable compounds, ideally out-of-distribution chemistry, and the retained set becomes trustworthy without retraining. We audit this recipe for a gradient-boosted fingerprint model on lipophilicity (regression) and BBBP (classification) under a Bemis-Murcko scaffold split, holding the fitted model and evaluation molecules fixed while varying only the calibration fold (18 replicates). (i) On lipophilicity, coverage depends on calibration provenance: calibrating on held-out training-scaffold molecules falls below nominal (PICP 0.841 at a 0.90 target), while novel-scaffold calibration recovers it (0.904); on BBBP, coverage stays above nominal in every regime. (ii) For split conformalized quantile regression (CQR) the conformal correction is a constant, so conformalization does not change what is abstained on. (iii) The abstention score (CQR width; effectively ensemble variance on BBBP) tracks local label variance more than chemical novelty: controlling for local label variance largely explains away the novelty association (partial Spearman 0.067 and 0.047). (iv) At a fixed assay budget, gating on interval width or ensemble variance never improves hit precision over ranking by point prediction, at the pre-specified operating point or across 20 exploratory settings; a plain novelty gate shows small, uncorrected gains at the smallest budgets. (v) In an exploratory analysis, exact conformal selection exceeds its false discovery rate target under training-scaffold calibration (0.303 at q=0.2) and is consistent with it under novel-scaffold calibration, contributing a data-driven batch size rather than a better ranking. Conformalization earns its keep through guarantees whose validity hinges on calibration provenance, not by choosing what to abstain on.

## A Utility-Mining-Driven Active Learning Approach for Analyzing Clickstream Sequences

DevFeed: [A Utility-Mining-Driven Active Learning Approach for Analyzing Clickstream Sequences](<https://devfeed.tech/articles/a-utility-mining-driven-active-learning-approach-for-analyzing-clickstream-sequences-108440.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2410.07282>)

Author: Danny Y. C. Wang, Lars Arne Jordanger, Jerry Chun-Wei Lin

Published: 2026-10-30T04:00:00Z

Content type: article

Language: en

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

Topics: [Predictive AI](<https://devfeed.tech/topics/predictive-ai.md>), [Data science and engineering](<https://devfeed.tech/topics/data-science-and-engineering.md>), [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>)

Tags: [behaviors](<https://devfeed.tech/tags/behaviors.md>), [commerce](<https://devfeed.tech/tags/commerce.md>), [cost](<https://devfeed.tech/tags/cost.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [data-processing](<https://devfeed.tech/tags/data-processing.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [high-quality-data](<https://devfeed.tech/tags/high-quality-data.md>), [labeling](<https://devfeed.tech/tags/labeling.md>), [learning](<https://devfeed.tech/tags/learning.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [mining](<https://devfeed.tech/tags/mining.md>), [model-training](<https://devfeed.tech/tags/model-training.md>), [modeling](<https://devfeed.tech/tags/modeling.md>), [performance](<https://devfeed.tech/tags/performance.md>), [prediction](<https://devfeed.tech/tags/prediction.md>), [sequences](<https://devfeed.tech/tags/sequences.md>), [settings](<https://devfeed.tech/tags/settings.md>), [shap](<https://devfeed.tech/tags/shap.md>), [strategy](<https://devfeed.tech/tags/strategy.md>), [utility](<https://devfeed.tech/tags/utility.md>), [values](<https://devfeed.tech/tags/values.md>)

### AI overview

HUSPM-SHAP uses SHAP values with high-utility sequential pattern mining to select e-commerce clickstream data for model training, aiming to reduce labeling needs while retaining predictive performance for purchase and non-purchase behavior. The abstract reports strong results across diverse scenarios and notes that settings for positive and negative SHAP values affect which patterns are mined. It does not provide experiment details or quantify the labeling reduction.

### Source excerpt

arXiv:2410.07282v2 Announce Type: replace Abstract: In rapidly evolving e-commerce industry, the capability of selecting high-quality data for model training is essential. This study introduces the High-Utility Sequential Pattern Mining using SHAP values (HUSPM-SHAP) model, a utility mining-based active learning strategy to tackle this challenge. We found that the parameter settings for positive and negative SHAP values impact the model's mining outcomes, introducing a key consideration into the active learning framework. Through extensive experiments aimed at predicting behaviors that do lead to purchases or not, the designed HUSPM-SHAP model demonstrates its superiority across diverse scenarios. The model's ability to mitigate labeling needs while maintaining high predictive performance is highlighted. Our findings demonstrate the model's capability to refine e-commerce data processing, steering towards more streamlined, cost-effective prediction modeling.

## TRIDENT-Seq: Profile-Matched Prefix-Future Contrast for Label-Efficient Payment Risk Modeling

DevFeed: [TRIDENT-Seq: Profile-Matched Prefix-Future Contrast for Label-Efficient Payment Risk Modeling](<https://devfeed.tech/articles/trident-seq-profile-matched-prefix-future-contrast-for-label-efficient-payment-risk-modeling-104072.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.37365>)

Author: Weiyu Han, Haoyu Jiang, Jie Sheng, Ruobing Zuo, Huilin He, Ze Zhao, Xiao Tan, Nairong Zhang, Shuai Chen, Tianyi Zhang

Published: 2026-10-28T04:00:00Z

Content type: article

Language: en

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

Topics: [AI Foundation Models](<https://devfeed.tech/topics/ai-foundation-models.md>), [AI Models](<https://devfeed.tech/topics/ai-models.md>), [Predictive AI](<https://devfeed.tech/topics/predictive-ai.md>), [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [all](<https://devfeed.tech/tags/all.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [based](<https://devfeed.tech/tags/based.md>), [behavior](<https://devfeed.tech/tags/behavior.md>), [budget](<https://devfeed.tech/tags/budget.md>), [business](<https://devfeed.tech/tags/business.md>), [but](<https://devfeed.tech/tags/but.md>), [can](<https://devfeed.tech/tags/can.md>), [chargeback](<https://devfeed.tech/tags/chargeback.md>), [continuation](<https://devfeed.tech/tags/continuation.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [deployment](<https://devfeed.tech/tags/deployment.md>), [dispersion](<https://devfeed.tech/tags/dispersion.md>), [easy](<https://devfeed.tech/tags/easy.md>), [future](<https://devfeed.tech/tags/future.md>), [modeling](<https://devfeed.tech/tags/modeling.md>), [payment](<https://devfeed.tech/tags/payment.md>), [profile](<https://devfeed.tech/tags/profile.md>), [risk](<https://devfeed.tech/tags/risk.md>)

### AI overview

TRIDENT-Seq uses Prefix-Future Contrast (PFC) to pretrain a transaction encoder: it matches a transaction-history prefix to a summary of later behavior among candidate histories with similar activity profiles, while activity prediction and dispersion regularization shape the encoder output. Across three payment businesses, it improves average precision and recall at a 0.5% intervention budget over architecture-matched training from scratch in all six settings. It also exceeds full-label scratch training in mean average precision and recall using 10% of Wallet labels and 25% of Shopping and SME labels; pretraining on another business's histories still helps. Replacing profile-matched candidates with random ones reduces recall from 24.62% to 5.64%. A deployment evaluation reports 37.7% lower amount-based chargeback rates and 37.9% lower overall loss rates than the previous XGBoost system at the same intervention budget.

### Source excerpt

arXiv:2610.37365v1 Announce Type: new Abstract: Ant International accumulates large volumes of transaction history across users and payment businesses. Confirmed fraud and chargeback outcomes, by contrast, are scarce, delayed, and specific to each business. A single encoder pretrained on unlabeled histories could serve all businesses, but only if pretraining captures the behavior that separates risk. Coarse activity differences can make contrastive pretraining on transaction logs too easy: transaction volume, amount, and timing can distinguish a history's continuation from arbitrary candidates without modeling how payment resources are used. We present TRIDENT-Seq, a transaction foundation model trained with Prefix-Future Contrast (PFC). PFC matches each prefix to a summary of its subsequent behavior among candidates with similar activity profiles; activity prediction and dispersion regularization act on the encoder output. Across three production payment businesses, the pretrained encoder improves average precision (AP) and recall at a 0.5% intervention budget over architecture-matched scratch training in all six business-budget settings. With a fixed validation set, it exceeds full-label scratch training in mean AP and recall using 10% of the training labels on Wallet and 25% on Shopping and SME. Pretraining only on another business's histories still helps. Replacing profile-matched candidates with random ones cuts recall from 24.62% to 5.64%. In Ant International's deployment evaluation, recorded amount-based chargeback and overall loss rates are 37.7% and 37.9% lower than those of the previous XGBoost-based system under the same intervention budget.

## AlphaMemo: Structured Search-Process Memory for Self-Evolving Alpha Mining Agents

DevFeed: [AlphaMemo: Structured Search-Process Memory for Self-Evolving Alpha Mining Agents](<https://devfeed.tech/articles/alphamemo-structured-search-process-memory-for-self-evolving-alpha-mining-agents-99428.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2606.20625>)

Author: Hang Yu, Zifan Zheng, Chun Gan, Jeff Z. Pan, Tongliang Liu, Zhiyong Wang, Fengxiang He

Published: 2026-10-26T04:00:00Z

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: [agentic workflows](<https://devfeed.tech/topics/agentic-workflows.md>), [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [agents](<https://devfeed.tech/tags/agents.md>), [alpha](<https://devfeed.tech/tags/alpha.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>), [memory](<https://devfeed.tech/tags/memory.md>), [mining](<https://devfeed.tech/tags/mining.md>), [process](<https://devfeed.tech/tags/process.md>), [search](<https://devfeed.tech/tags/search.md>), [structured](<https://devfeed.tech/tags/structured.md>)

### AI overview

AlphaMemo addresses redundant discoveries and overfitting in alpha mining by storing evidence about which code-edit motifs succeed or fail in specific parent-factor contexts, rather than retaining only final factors or complete search histories. It extracts motifs from differences between factor expression syntax trees, adds confidence-gated residual memory to a search-ledger prior, and uses asymmetric veto control to suppress high-confidence failure patterns. Experiments on CSI 500 and S&P 500 report improved out-of-sample performance and discovery efficiency under a fixed budget; ablations support the contributions of the memory components. The abstract does not provide numerical results or further experimental details.

### Source excerpt

arXiv:2606.20625v2 Announce Type: replace-cross Abstract: LLM agents are promising for alpha mining via combining financial priors, symbolic reasoning, executable factor generation, and feedback-driven refinement. Yet, they face a combinatorial search space, noisy non-stationary feedback, redundant discoveries, and overfitting risks from naively reusing past successes. To address these challenges, we propose AlphaMemo, a self-evolving alpha mining agent with Structured Search-Process Memory. Rather than memorizing only final factors or full trajectories, AlphaMemo records reusable evidence about which edit motifs work or fail under specific parent-factor contexts. It extracts motifs from Abstract Syntax Tree (AST) differences, applies confidence-gated residual memory on top of a search-ledger prior, and uses asymmetric veto control to suppress high-confidence failure patterns. Experiments on CSI 500 and S&P 500 show improved out-of-sample performance and fixed-budget discovery efficiency, with ablations validating the roles of residual learning, confidence gating, AST-diff motifs, and veto memory. Code is at https://github.com/jarrettyu/sspm.

## Lineage-Aware Search Control for LLM-Based Evolutionary Factor Mining

DevFeed: [Lineage-Aware Search Control for LLM-Based Evolutionary Factor Mining](<https://devfeed.tech/articles/lineage-aware-search-control-for-llm-based-evolutionary-factor-mining-94973.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.29663>)

Author: Xintong Liu, Chenkai Zhang, Yanzhao Ma, Peng Yang

Published: 2026-10-22T04:00:00Z

Content type: article

Language: en

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

Topics: [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [Hyperparameter optimization](<https://devfeed.tech/topics/hyperparameter-optimization.md>)

Tags: [alpha](<https://devfeed.tech/tags/alpha.md>), [budget](<https://devfeed.tech/tags/budget.md>), [control](<https://devfeed.tech/tags/control.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [data](<https://devfeed.tech/tags/data.md>), [discovery](<https://devfeed.tech/tags/discovery.md>), [diversity](<https://devfeed.tech/tags/diversity.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [evolution](<https://devfeed.tech/tags/evolution.md>), [finance](<https://devfeed.tech/tags/finance.md>), [generation](<https://devfeed.tech/tags/generation.md>), [large-language-models](<https://devfeed.tech/tags/large-language-models.md>), [llm](<https://devfeed.tech/tags/llm.md>), [markets](<https://devfeed.tech/tags/markets.md>), [mining](<https://devfeed.tech/tags/mining.md>), [portfolio](<https://devfeed.tech/tags/portfolio.md>), [q-fin-cp](<https://devfeed.tech/tags/q-fin-cp.md>), [quality](<https://devfeed.tech/tags/quality.md>), [quantitative-finance](<https://devfeed.tech/tags/quantitative-finance.md>), [search](<https://devfeed.tech/tags/search.md>), [selection](<https://devfeed.tech/tags/selection.md>), [semantic](<https://devfeed.tech/tags/semantic.md>)

### AI overview

LineageEvo uses evolutionary descent as a control unit for LLM-based formulaic alpha factor mining. It guides generation with lineage-specific, operator-level priors combined with global priors, and allocates search budget across lineages to limit domination by a few branches while balancing factor quality and lineage concentration. On CSI 500 and S&P 500, it reports the strongest predictive performance; LSPS yields a more balanced lineage distribution and structurally broader factor pool. The abstract also says the evaluation covers portfolio performance, search efficiency, and lineage- and expression-level diversity, but provides no detailed metrics.

### Source excerpt

arXiv:2610.29663v1 Announce Type: cross Abstract: Formulaic alpha factor mining searches for symbolic expressions that transform market data into predictive signals. Recent LLM-based evolutionary factor mining methods use large language models as semantic variation operators. Under the conventional LLM-EA paradigm, stochastic variation and fitness-based selection can leave the search weakly directed and concentrated around a narrow set of structurally similar factors. We argue that these limitations arise because current methods mainly organize search at the individual and population levels, leaving lineage-level control underused. A lineage connects successive factor variants through evolutionary descent. We propose LineageEvo, which uses lineage in two coupled forms of search control: (1) lineage-conditioned LLM variation, which distills Lineage-Scoped Search Experience into operator-specific priors and fuses them with global operator priors, providing persistent, lineage-relevant guidance for LLM generation; and (2) lineage-stratified search allocation, which balances factor quality against lineage concentration by distributing the search budget across lineages during evolutionary search. Together, these mechanisms make semantic variation more direction aware and prevent branches from dominating the evolution, thereby promoting the discovery of predictive and diverse factors. We evaluate LineageEvo on CSI 500 and S&P 500 in terms of factor predictiveness, portfolio performance, search efficiency, and lineage- and expression-level diversity. LineageEvo achieves the strongest predictive performance on both markets, while LSPS produces a more balanced lineage distribution and a structurally broader factor pool. The results support lineage as a useful intermediate unit for controlling LLM-based evolutionary factor search.

## Local and Global Recursive Ridge Regression for Online Residual Correction of Frozen Multivariate Forecasters under Label Delay

DevFeed: [Local and Global Recursive Ridge Regression for Online Residual Correction of Frozen Multivariate Forecasters under Label Delay](<https://devfeed.tech/articles/local-and-global-recursive-ridge-regression-for-online-residual-correction-of-frozen-multivariate-forecasters-under-label-delay-94963.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.29604>)

Author: Takumi Fujimoto, Hiroaki Nishi

Published: 2026-10-22T04:00:00Z

Content type: article

Language: en

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

Topics: [linear-regression](<https://devfeed.tech/topics/linear-regression.md>), [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>)

Tags: [channel](<https://devfeed.tech/tags/channel.md>), [cosine](<https://devfeed.tech/tags/cosine.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [datasets](<https://devfeed.tech/tags/datasets.md>), [delay](<https://devfeed.tech/tags/delay.md>), [electricity](<https://devfeed.tech/tags/electricity.md>), [elf](<https://devfeed.tech/tags/elf.md>), [error](<https://devfeed.tech/tags/error.md>), [forecast](<https://devfeed.tech/tags/forecast.md>), [global](<https://devfeed.tech/tags/global.md>), [local](<https://devfeed.tech/tags/local.md>), [online](<https://devfeed.tech/tags/online.md>), [reduction](<https://devfeed.tech/tags/reduction.md>), [regression](<https://devfeed.tech/tags/regression.md>), [state](<https://devfeed.tech/tags/state.md>), [statistics](<https://devfeed.tech/tags/statistics.md>), [sufficient](<https://devfeed.tech/tags/sufficient.md>), [times](<https://devfeed.tech/tags/times.md>), [traffic](<https://devfeed.tech/tags/traffic.md>), [weather](<https://devfeed.tech/tags/weather.md>)

### AI overview

DRiC corrects a frozen multi-horizon forecaster online by recursively fitting ridge regressions on leading DCT coefficients of forecast and input blocks to predict residual blocks, accommodating the delay before each block's labels arrive. DRiC-L keeps a regression per channel; DRiC-G pools channels, so its statistics do not grow with channel count. A guard skips updates for unusually large residual blocks and clips corrections. Across 126 evaluated conditions, DRiC-L8 reduced MSE by 14.55% on average versus 12.98% for ELF, while retaining 15-17 times less state; ELF was more accurate on Traffic. DRiC-G used 14.6 kB on Electricity and 18.9 kB on Traffic and was the most accurate method at or below its state, though 5-7 comparators using at least 10 times more state performed better. On unfilled Weather data, the guard mattered: without it, DRiC-L8's MSE averaged 8 times the frozen forecast's; with it, MSE fell 7.94%.

### Source excerpt

arXiv:2610.29604v1 Announce Type: new Abstract: A frozen multi-horizon forecaster can be corrected online from its own residuals. However, under label delay the target of a block arrives only after the whole horizon has passed, and the corrector's retained state grows with the number of channels. Aiming to combine a small retained state with a large error reduction, we propose DRiC (DCT Ridge Corrector), which fits a ridge regression of the leading discrete cosine transform (DCT) coefficients of the next residual block on those of the frozen forecast and of the latest input block, with sufficient statistics updated recursively. Its local variant, DRiC-L, keeps one such regression per channel; its global variant, DRiC-G, pools all channels into one regression, whose statistics do not depend on the channel count. A guard tracks a running scale of the frozen forecast's residual for each channel; it skips updates on blocks far outside that scale and clips the issued correction. We evaluate on seven datasets (four ETT series, Weather with its missing-value codes filled, Electricity, and Traffic) with DLinear, PatchTST, and iTransformer backbones, against 10 naive, gradient-trained, and closed-form methods. DRiC-L with eight DCT coefficients (DRiC-L8) reduces the mean squared error (MSE) by 14.55% on average over 126 conditions, against 12.98% for ELF, the most accurate comparator, while retaining 15-17 times less state, although ELF is more accurate on Traffic. DRiC-G retains 14.6 kB on the 321-channel Electricity data and 18.9 kB on the 862-channel Traffic data and is the most accurate method at or below its state on both. However, 5 to 7 of the 10 comparators, all with at least 10 times its state, are more accurate there. On the unfilled Weather file, DRiC-L8 without the guard has on average 8.0 times the frozen forecast's MSE, whereas with the guard it reduces the MSE by 7.94%. Code: https://github.com/keiotakmin/dric.

## Multi-Label Sentiment Analysis on 100 Languages with Dynamic Weighting for Label Imbalance

DevFeed: [Multi-Label Sentiment Analysis on 100 Languages with Dynamic Weighting for Label Imbalance](<https://devfeed.tech/articles/multi-label-sentiment-analysis-on-100-languages-with-dynamic-weighting-for-label-imbalance-94656.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2008.11573>)

Author: Selim F. Yilmaz, E. Batuhan Kaynak, Aykut Ko\\c{c}, Hamdi Dibeklio\\u{g}lu, Suleyman S. Kozat

Published: 2026-10-22T04:00:00Z

Content type: article

Language: en

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

Topics: [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>), [Internationalization (i18n)](<https://devfeed.tech/topics/i18n.md>)

Tags: [applications](<https://devfeed.tech/tags/applications.md>), [attention](<https://devfeed.tech/tags/attention.md>), [best](<https://devfeed.tech/tags/best.md>), [class](<https://devfeed.tech/tags/class.md>), [code](<https://devfeed.tech/tags/code.md>), [competition](<https://devfeed.tech/tags/competition.md>), [complexity](<https://devfeed.tech/tags/complexity.md>), [contribution](<https://devfeed.tech/tags/contribution.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>), [experiments](<https://devfeed.tech/tags/experiments.md>), [framework](<https://devfeed.tech/tags/framework.md>), [languages](<https://devfeed.tech/tags/languages.md>), [market-research](<https://devfeed.tech/tags/market-research.md>), [metrics](<https://devfeed.tech/tags/metrics.md>), [model](<https://devfeed.tech/tags/model.md>), [novel](<https://devfeed.tech/tags/novel.md>), [object](<https://devfeed.tech/tags/object.md>), [performance](<https://devfeed.tech/tags/performance.md>), [politics](<https://devfeed.tech/tags/politics.md>), [research](<https://devfeed.tech/tags/research.md>), [stat-ml](<https://devfeed.tech/tags/stat-ml.md>)

### AI overview

The paper presents a multi-label cross-lingual sentiment model based on Plutchik's wheel of emotions. Its training method dynamically weights classes to address label imbalance, adapts focal loss to emphasize harder examples, and selects class-specific thresholds to maximize macro-F1 in linear time. The authors report state-of-the-art results on 7 of 9 metrics across three evaluated languages, using one model, against common baselines and leading SemEval methods. They share code for a model that can perform sentiment analysis in 100 languages; the abstract's reported performance comparison covers only three languages.

### Source excerpt

arXiv:2008.11573v2 Announce Type: replace Abstract: We investigate cross-lingual sentiment analysis, which has attracted significant attention due to its applications in various areas including market research, politics and social sciences. In particular, we introduce a sentiment analysis framework in multi-label setting as it obeys Plutchik's wheel of emotions. We introduce a novel dynamic weighting method that balances the contribution from each class during training, unlike previous static weighting methods that assign non-changing weights based on their class frequency. Moreover, we adapt the focal loss that favors harder instances from single-label object recognition literature to our multi-label setting. Furthermore, we derive a method to choose optimal class-specific thresholds that maximize the macro-f1 score in linear time complexity. Through an extensive set of experiments, we show that our method obtains the state-of-the-art performance in 7 of 9 metrics in 3 different languages using a single model compared to the common baselines and the best-performing methods in the SemEval competition. We publicly share our code for our model, which can perform sentiment analysis in 100 languages, to facilitate further research.

[Next page](<https://devfeed.tech/topics/data-mining-modeling.md?cursor=WyIyMDI2LTEwLTIyVDA0OjAwOjAwKzAwOjAwIiwgIjQ3NDIwMmFmLWYyZTYtNDNlZC1hMzkzLTJiNTQ4YmRiMzQ4NiJd>)