# Decoding

Decoding is the process of recovering original data from an encoded signal or representation, often despite noise or interference.

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

## Decoding base64 email attachments: a speed ranking of programming languages

DevFeed: [Decoding base64 email attachments: a speed ranking of programming languages](<https://devfeed.tech/articles/decoding-base64-email-attachments-a-speed-ranking-of-programming-languages-146498.md>)

Original publisher: [Read original article](<https://lemire.me/blog/2026/11/20/decoding-base64-email-attachments/>)

Author: Daniel Lemire

Published: 2026-11-20T12:37:52Z

Content type: news

Language: en

Sources: [Daniel Lemire](<https://devfeed.tech/sources/daniel-lemire.md>)

Topics: [Decoding](<https://devfeed.tech/topics/decoding.md>), [Programming](<https://devfeed.tech/topics/programming.md>), [Programming Languages / Compilers](<https://devfeed.tech/topics/programming-languages-compilers.md>), [email](<https://devfeed.tech/topics/email.md>)

Tags: [base64](<https://devfeed.tech/tags/base64.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [decoding](<https://devfeed.tech/tags/decoding.md>), [email](<https://devfeed.tech/tags/email.md>), [languages](<https://devfeed.tech/tags/languages.md>), [mime](<https://devfeed.tech/tags/mime.md>), [php](<https://devfeed.tech/tags/php.md>), [programming](<https://devfeed.tech/tags/programming.md>), [programming-languages](<https://devfeed.tech/tags/programming-languages.md>), [pull-request](<https://devfeed.tech/tags/pull-request.md>), [ranking](<https://devfeed.tech/tags/ranking.md>), [speed](<https://devfeed.tech/tags/speed.md>)

### AI overview

A benchmark of MIME-style base64 decoding found a thirtyfold gap between the fastest and slowest tested systems. It used one megabyte of random data encoded in 76-character lines with CRLF, verified decoded bytes, and measured median throughput across repeated runs on one pinned core of an Intel Xeon Gold 6548N. The excerpt says the top six systems use simdutf, which removes line endings in its vector loop and compacts affected blocks into a small buffer instead of falling back to a slow path. PHP's current release measured 0.61 GB/s on this input despite exceeding 20 GB/s on uninterrupted base64; a proposed simdutf port reportedly made MIME decoding 28 times faster, moving PHP to second place. The excerpt does not provide the other systems' individual results.

### Source excerpt

Email attachments are base64. So are cryptographic PEM certificates. In both cases the text is broken into lines, 76 characters for email and 64 for PEM. Skipping line endings sounds trivial. Yet many fast decoders only work on uninterrupted base64. As soon as they meet a line break, they fall back to a slow path. ... Continue reading Decoding base64 email attachments: a speed ranking of programming languages

## AscMLA: Unlocking FP4 Efficiency for Long-Context MLA Decoding through Asymmetric Scaling

DevFeed: [AscMLA: Unlocking FP4 Efficiency for Long-Context MLA Decoding through Asymmetric Scaling](<https://devfeed.tech/articles/ascmla-unlocking-fp4-efficiency-for-long-context-mla-decoding-through-asymmetric-scaling-146129.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.26106>)

Author: Zechun Zhou, Zhiyong Zheng, Xiaomin Zhang, Lin Ye, Han Li, Jingwei Sun, Guangzhong Sun, Jing Li

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

Content type: article

Language: en

Sources: [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>)

Topics: [GPU optimization](<https://devfeed.tech/topics/gpu-optimization.md>), [NVFP4](<https://devfeed.tech/topics/nvfp4.md>), [Decoding](<https://devfeed.tech/topics/decoding.md>)

Tags: [attention](<https://devfeed.tech/tags/attention.md>), [cache](<https://devfeed.tech/tags/cache.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [decoding](<https://devfeed.tech/tags/decoding.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [fp4](<https://devfeed.tech/tags/fp4.md>), [long-context](<https://devfeed.tech/tags/long-context.md>), [memory](<https://devfeed.tech/tags/memory.md>), [mla](<https://devfeed.tech/tags/mla.md>), [scaling](<https://devfeed.tech/tags/scaling.md>), [storage](<https://devfeed.tech/tags/storage.md>)

### AI overview

AscMLA addresses the mismatch between K/V scaling requirements when FP4 is applied to MLA's shared latent. It combines asymmetric K/V scaling with pipelined scale delivery to process one FP4 payload directly, avoiding the dequantization or separate encodings used by existing approaches, which add execution or storage overhead. The abstract reports 38.2% less cache memory than FP8 and geometric-mean decoding speedups of 1.35x over TRT-LLM's FP8 kernels and 1.26x over CAKE's NVFP4 MLA kernels.

### Source excerpt

arXiv:2611.26106v1 Announce Type: new Abstract: Long-context decoding repeatedly reads a growing KV cache. FP4 reduces cache storage and traffic but needs block scaling to handle its limited dynamic range. Applying FP4 to multi-head latent attention (MLA) requires reconciling different K/V scaling requirements on a shared latent. Existing approaches address this mismatch through dequantization or separate encodings, adding execution or storage overhead. We present AscMLA, which combines asymmetric K/V scaling with pipelined scale delivery to process a single FP4 payload directly. AscMLA uses 38.2% less cache memory than FP8 and achieves geometric-mean speedups of 1.35x over TRT-LLM's FP8 and 1.26x over CAKE's NVFP4 MLA kernels.

## GroundAnything: Reconciling Parallel Decoding with Precise Visual Grounding at Flash Speed

DevFeed: [GroundAnything: Reconciling Parallel Decoding with Precise Visual Grounding at Flash Speed](<https://devfeed.tech/articles/groundanything-reconciling-parallel-decoding-with-precise-visual-grounding-at-flash-speed-144950.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2609.39600>)

Author: Qize Yu, Lianrui Fan, Bowen Ping, Xini Ding, Zetian Song, Junbo Niu, Kaixuan Wang, Tianxing Chen, Yue Chen, Minghua He, Yuran Wang, Jie Huang, Haojun Zhang, Min Chen, Hao Li, Wenxuan Song, Ruihai Wu,

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

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: [AI Foundation Models](<https://devfeed.tech/topics/ai-foundation-models.md>), [Decoding](<https://devfeed.tech/topics/decoding.md>), [Fine-tuning](<https://devfeed.tech/topics/fine-tuning.md>), [grpo](<https://devfeed.tech/topics/grpo.md>), [post-training](<https://devfeed.tech/topics/post-training.md>), [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [also](<https://devfeed.tech/tags/also.md>), [ar](<https://devfeed.tech/tags/ar.md>), [art](<https://devfeed.tech/tags/art.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [astra](<https://devfeed.tech/tags/astra.md>), [autoregressive](<https://devfeed.tech/tags/autoregressive.md>), [based](<https://devfeed.tech/tags/based.md>), [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [coco](<https://devfeed.tech/tags/coco.md>), [combines](<https://devfeed.tech/tags/combines.md>), [conversion](<https://devfeed.tech/tags/conversion.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [data](<https://devfeed.tech/tags/data.md>), [datasets](<https://devfeed.tech/tags/datasets.md>), [decoding](<https://devfeed.tech/tags/decoding.md>), [flash](<https://devfeed.tech/tags/flash.md>), [grounding](<https://devfeed.tech/tags/grounding.md>), [parallel](<https://devfeed.tech/tags/parallel.md>), [speed](<https://devfeed.tech/tags/speed.md>), [with](<https://devfeed.tech/tags/with.md>)

### AI overview

GroundAnything is a 4B-parameter visual grounding model that uses blockwise diffusion denoising to generate and refine spatial hypotheses in parallel, addressing the sequential latency of autoregressive decoding. Its training combines public-data and dedicated-data-engine pretraining, conversion from autoregressive models with joint objectives, supervised fine-tuning, and GRPO reinforcement post-training. Across 30 benchmarks, its autoregressive variant scores 72.42%, compared with 71.35% for GPT-6 Astra among similarly sized models. With entropy-guided decoding, GroundAnything averages 61.75%, versus 53.32% for the fast MTP-based LocateAnything. An optional self-speculative mode reports a 4.51x speedup over the autoregressive counterpart, with a 0.74-point drop in COCO F1mIoU. The excerpt also says inference optimizations yield practical speedups, without detailing them.

### Source excerpt

arXiv:2609.39600v3 Announce Type: replace-cross Abstract: Autoregressive (AR) grounding models serialize spatial predictions, introducing sequential latency and imposing a causal order on output tokens. We view grounding as visual evidence extraction: objects, locations, and spatial relations are jointly constrained by the image and query, yet their dependencies do not imply an intrinsic left-to-right generation order. This distinction makes bidirectional diffusion a natural fit, allowing spatial hypotheses to emerge in parallel and be jointly refined through iterative denoising. We introduce GroundAnything, a 4B-parameter grounding foundation model that reconciles fast parallel decoding with precise localization through blockwise denoising. Training combines grounding pretraining from public datasets and dedicated data engines, direct AR-to-diffusion conversion with joint AR and diffusion objectives, supervised fine-tuning, and GRPO-based reinforcement post-training. Across 30 grounding benchmarks, our autoregressive variant, GroundAnything-VLM, establishes a new overall state of the art among similarly sized models at 72.42%, remaining competitive with GPT-6 Astra (71.35%). With entropy-guided decoding, GroundAnything also surpasses the prior state of the art at this scale, averaging 61.75% versus 53.32% for the fast MTP-based LocateAnything model. We further explore decoding strategies, showing that an optional self-speculative mode achieves a 4.51$\times$ speedup over the AR counterpart with a 0.74 percentage-point drop in COCO F1mIoU. Infrastructure experiments show that progressive inference optimizations translate parallel decoding into practical speedups. These gains support efficient visual grounding in latency-sensitive real-world systems.

## Raptor: Rejection-aware Adaptive Tree based on Speculative Decoding

DevFeed: [Raptor: Rejection-aware Adaptive Tree based on Speculative Decoding](<https://devfeed.tech/articles/raptor-rejection-aware-adaptive-tree-based-on-speculative-decoding-145514.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.25436>)

Author: Qishu Hu, Haozhao Wang, Zicong Hong, Yufeng Zhan, Yuanqing Xia

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

Content type: article

Language: en

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

Topics: [Decoding](<https://devfeed.tech/topics/decoding.md>), [Model Optimizer](<https://devfeed.tech/topics/model-optimizer.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>)

Tags: [acceptance](<https://devfeed.tech/tags/acceptance.md>), [adaptive](<https://devfeed.tech/tags/adaptive.md>), [autoregressive](<https://devfeed.tech/tags/autoregressive.md>), [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [budget](<https://devfeed.tech/tags/budget.md>), [candidate](<https://devfeed.tech/tags/candidate.md>), [control](<https://devfeed.tech/tags/control.md>), [cost](<https://devfeed.tech/tags/cost.md>), [coverage](<https://devfeed.tech/tags/coverage.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [decoding](<https://devfeed.tech/tags/decoding.md>), [deep](<https://devfeed.tech/tags/deep.md>), [effectiveness](<https://devfeed.tech/tags/effectiveness.md>), [effort](<https://devfeed.tech/tags/effort.md>), [execution](<https://devfeed.tech/tags/execution.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [feedback](<https://devfeed.tech/tags/feedback.md>), [framework](<https://devfeed.tech/tags/framework.md>), [frontier](<https://devfeed.tech/tags/frontier.md>), [raptor](<https://devfeed.tech/tags/raptor.md>), [speculative-decoding](<https://devfeed.tech/tags/speculative-decoding.md>)

### AI overview

Raptor is a training-free framework for adapting tree-structured speculative decoding in vision-language models, where long visual contexts make speculative computation costly. It uses complementary rejection signals to distinguish weak candidate coverage from redundant deep expansion, since acceptance length alone cannot identify which adjustment is needed. Those signals guide bounded, momentum-smoothed changes to tree depth, branching width, and drafting budget, while execution control curbs unnecessary drafting and restores expansion progressively. Experiments with two VLMs across seven multimodal benchmarks report up to 2.39x average speedup over autoregressive decoding.

### Source excerpt

arXiv:2611.25436v1 Announce Type: cross Abstract: Tree-structured speculative decoding accelerates vision-language models (VLMs) by verifying multiple candidate continuations in parallel. However, long visual contexts increase the cost of speculative computation, making candidate coverage and redundant expansion important considerations beyond tree topology adaptation. We propose Raptor, a training-free framework that uses rejection feedback to jointly regulate candidate coverage and drafting effort. Our key insight is that acceptance length alone cannot distinguish insufficient candidate coverage from excessive deep expansion, and these conditions require different allocation decisions. Raptor therefore extracts complementary rejection-aware signals to characterize speculative progress, frontier coverage, and redundant expansion. These signals jointly guide bounded, momentum-smoothed updates of tree depth, branching width, and drafting budget, together with execution control that limits unnecessary drafting and progressively restores expansion when appropriate. Extensive experiments with two VLMs across seven multimodal benchmarks demonstrate the effectiveness of Raptor, achieving up to 2.39x average speedup over autoregressive decoding.

## Supportive Token Revealing for Fast Diffusion Language Model Decoding

DevFeed: [Supportive Token Revealing for Fast Diffusion Language Model Decoding](<https://devfeed.tech/articles/supportive-token-revealing-for-fast-diffusion-language-model-decoding-144901.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2606.04236>)

Author: Giries Abu Ayoub, Mario Barbara, Llu\\'is Pastor-P\\'erez, Tanja Bien, Aneesh Barthakur, Alaa Maalouf, Loay Mualem

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

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: [Language models](<https://devfeed.tech/topics/language-models.md>), [LLM optimization](<https://devfeed.tech/topics/llm-optimization.md>), [Decoding](<https://devfeed.tech/topics/decoding.md>)

Tags: [attention](<https://devfeed.tech/tags/attention.md>), [axon](<https://devfeed.tech/tags/axon.md>), [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [code](<https://devfeed.tech/tags/code.md>), [code-generation](<https://devfeed.tech/tags/code-generation.md>), [confidence](<https://devfeed.tech/tags/confidence.md>), [context](<https://devfeed.tech/tags/context.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>), [decode](<https://devfeed.tech/tags/decode.md>), [decoder](<https://devfeed.tech/tags/decoder.md>), [decoding](<https://devfeed.tech/tags/decoding.md>), [denoising](<https://devfeed.tech/tags/denoising.md>), [dependency](<https://devfeed.tech/tags/dependency.md>), [diffusion](<https://devfeed.tech/tags/diffusion.md>), [evaluations](<https://devfeed.tech/tags/evaluations.md>), [existing](<https://devfeed.tech/tags/existing.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [fast](<https://devfeed.tech/tags/fast.md>), [language](<https://devfeed.tech/tags/language.md>), [language-model](<https://devfeed.tech/tags/language-model.md>), [token](<https://devfeed.tech/tags/token.md>)

### AI overview

AXON is a training-free add-on for parallel diffusion language model decoders that aims to make remaining uncertain tokens easier to denoise. It monitors those tokens and, when they appear to need more context, uses attention, uncertainty, and confidence signals to select confident masked tokens as anchors for later denoising. This shifts the reveal criterion from which token is safest to reveal to which reveal would best support other tokens, while leaving the base decoder in place. On reasoning and code-generation benchmarks across multiple diffusion models, the abstract reports improved quality-latency trade-offs, often with fewer function evaluations while maintaining or improving accuracy.

### Source excerpt

arXiv:2606.04236v2 Announce Type: replace-cross Abstract: Discrete diffusion language models can generate text efficiently by updating multiple masked positions in parallel, but this parallelism introduces a quality-latency trade-off. Aggressive decoding may commit mutually dependent tokens too early, while conservative decoding requires many denoising steps. Existing methods address this tension by deciding which tokens are safe to reveal using confidence or dependency criteria. However, avoiding unsafe commits does not necessarily make the remaining masked sequence easy to decode, since uncertain tokens may depend on masked tokens, creating a bottleneck for denoising steps. We propose AXON, a training-free module that can be added on top of existing parallel decoding strategies for diffusion language models. Rather than replacing the base decoder, AXON monitors the remaining uncertain masked tokens and intervenes only when their current state suggests that additional context is needed. It then shifts the criterion from which tokens are safest to reveal to which confident reveals would best support later denoising. AXON selects anchors, confident masked tokens that uncertain positions attend to, using attention, uncertainty, and confidence signals. Experiments on reasoning and code-generation benchmarks across multiple diffusion language models show that AXON improves the quality-latency trade-off of existing parallel decoders, often reducing the number of function evaluations while maintaining or improving accuracy.

## LoopCache: Accelerating Long-Context Self-Speculative Decoding for Looped Transformers

DevFeed: [LoopCache: Accelerating Long-Context Self-Speculative Decoding for Looped Transformers](<https://devfeed.tech/articles/loopcache-accelerating-long-context-self-speculative-decoding-for-looped-transformers-143503.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.24301>)

Author: Zizhuo Fu, Haochen Huang, Shengxuan Qiu, Linye Wei, Ziyu Tang, Runsheng Wang, Meng Li

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

Content type: article

Language: en

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

Topics: [Model Optimizer](<https://devfeed.tech/topics/model-optimizer.md>), [Transformers](<https://devfeed.tech/topics/transformers.md>), [Decoding](<https://devfeed.tech/topics/decoding.md>)

Tags: [cache](<https://devfeed.tech/tags/cache.md>), [confidence](<https://devfeed.tech/tags/confidence.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [decoding](<https://devfeed.tech/tags/decoding.md>), [kv-cache](<https://devfeed.tech/tags/kv-cache.md>), [long-context](<https://devfeed.tech/tags/long-context.md>), [speculative-decoding](<https://devfeed.tech/tags/speculative-decoding.md>), [tokens](<https://devfeed.tech/tags/tokens.md>), [transformers](<https://devfeed.tech/tags/transformers.md>)

### AI overview

LoopCache targets two bottlenecks in wave-style self-speculative decoding for long-context Looped Transformers: fixed-loop drafts can become stale as predictions change, and batching tokens across loops can load up to R times the KV cache per step. It dynamically updates drafts using output confidence and rotates loop-specific KV caches so each step accesses one loop's cache while batching at the nearest matching cycle for lossless verification. It also retains independent caches for selected loop-sensitive shallow layers to improve acceptance rates with negligible overhead. The abstract reports average speedups of 10.25x over WFD and 14.56x over LoopSpec.

### Source excerpt

arXiv:2611.24301v1 Announce Type: cross Abstract: Looped Transformers achieve large effective depth with compact models by repeatedly applying a weight-shared block across R recurrent steps, each maintaining a loop-specific key-value (KV) cache. To accelerate sequential loop execution, recent wave-style self-speculative methods like WaveFront Decoding (WFD) and LoopSpec pipeline draft and target tokens across loops in a single step. However, they face two critical limitations. First, they draft only at fixed loops and ignore intermediate prediction changes, wasting computation on stale trajectories. Second, batching tokens from different loops loads up to Rx the KV cache per step, severely amplifying memory traffic in long-context scenarios. To address these challenges, we propose LoopCache, an efficient self-speculative decoding framework for long-context Looped Transformers. Based on the correlation between prediction switches and output confidence, LoopCache introduces confidence-guided dynamic draft updating to refine drafts in flight across all intermediate loops. Furthermore, LoopCache employs a cyclic loop KV cache strategy that rotates through loop-specific KV caches, restricting per-step KV traffic to a single loop while batching tokens at the nearest matching cycle for lossless verification. By also selectively retaining independent KV caches for a few loop-sensitive shallow layers, LoopCache improves acceptance rates with negligible overhead. Experiments show that LoopCache achieves average 10.25x and 14.56x speedup over WFD and LoopSpec.

## GALA: Gallery-Level Alignment for Real-Life Vision Decoding from EEG

DevFeed: [GALA: Gallery-Level Alignment for Real-Life Vision Decoding from EEG](<https://devfeed.tech/articles/gala-gallery-level-alignment-for-real-life-vision-decoding-from-eeg-142766.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.24206>)

Author: Philippe Forest, Giulia Lioi, Mai Pham Quyen, Alix Lamouroux, Javen Shi, Adrien Merlini

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

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: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [bci](<https://devfeed.tech/topics/bci.md>), [Decoding](<https://devfeed.tech/topics/decoding.md>)

Tags: [alignment](<https://devfeed.tech/tags/alignment.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [decoding](<https://devfeed.tech/tags/decoding.md>), [eeg](<https://devfeed.tech/tags/eeg.md>), [gallery](<https://devfeed.tech/tags/gallery.md>), [vision](<https://devfeed.tech/tags/vision.md>)

### AI overview

GALA improves EEG vision decoding for a new user without labeled calibration data by aligning the user's accumulating unlabeled EEG embeddings to a known gallery of candidate images. It uses shrinkage whitening, hubness correction, and Sinkhorn-Procrustes alignment; estimated query frequencies are updated after each query, which is decoded before being added to calibration. On THINGS-EEG-2 with 200 images, top-1 accuracy rose from 28% to 67% over 53 minutes of use. Across 60 tested settings, GALA was the most accurate label-free method in each, averaging 6.6 percentage points above the strongest competitor at the end of the stream. The evidence is from the abstract and does not establish performance beyond the described evaluations.

### Source excerpt

arXiv:2611.24206v1 Announce Type: new Abstract: Neural decoding of visual experience from electroencephalography (EEG) has achieved strong performance in intra-subject settings, but current methods fail to generalize across individuals: on the standard 200-way benchmark THINGS-EEG-2, top-1 accuracy drops from above 80% within subjects to about 32% for an unseen subject. Adapting a pre-trained encoder to new subjects and sessions typically requires labeled data, a bottleneck whose collection time severely limits practical device deployment. In a practical Brain-Computer Interface (BCI), however, the possible outcomes are known in advance and form a gallery of candidates, and the unlabeled EEG of the user accumulates as soon as the device is used. We propose a new setting, incremental transductive calibration, in which the decoder is calibrated regularly on all the unlabeled EEG received so far and on the known gallery, without any labels. This setting is closer to the conditions of a deployed BCI than the zero-shot and batch evaluations in current use. We evaluate it with a protocol in which queries arrive one at a time, with unknown and possibly unequal frequencies, and each query is decoded before it is used for calibration. In this setting, we introduce GALA : GAllery-Level Alignment, a label-free calibration method which registers the unlabeled EEG embeddings onto the gallery through shrinkage whitening, hubness correction and a Sinkhorn--Procrustes alignment with query frequencies estimated from the buffer, updated after every query. On THINGS-EEG-2, with a gallery of 200 images, it raises the top-1 accuracy on a new subject from 28% to 67% over 53 minutes of use. Across 60 evaluation settings that vary the gallery size, the query frequencies and the number of repetitions averaged per query, GALA is the most accurate label-free method in every setting, 6.6 points above the best competing method on average at the end of the stream.

## Sightline: Long-Window Speculative Decoding for Fast SVG Generation

DevFeed: [Sightline: Long-Window Speculative Decoding for Fast SVG Generation](<https://devfeed.tech/articles/sightline-long-window-speculative-decoding-for-fast-svg-generation-143448.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.23807>)

Author: Yanchao Li, Ben Gao, Jiaqing Xie, Yusu Zhang, Beilun Wang, Yanbo Wang, Xiaohua Xu, Nanqing Dong, Tianfan Fu, Zhongyu Huang

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

Content type: article

Language: en

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

Topics: [SVG](<https://devfeed.tech/topics/svg.md>), [Decoding](<https://devfeed.tech/topics/decoding.md>), [Image](<https://devfeed.tech/topics/image.md>), [lottie](<https://devfeed.tech/topics/lottie.md>)

Tags: [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [code](<https://devfeed.tech/tags/code.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [decoder](<https://devfeed.tech/tags/decoder.md>), [decoding](<https://devfeed.tech/tags/decoding.md>), [draft](<https://devfeed.tech/tags/draft.md>), [generation](<https://devfeed.tech/tags/generation.md>), [horizon](<https://devfeed.tech/tags/horizon.md>), [json](<https://devfeed.tech/tags/json.md>), [loops](<https://devfeed.tech/tags/loops.md>), [lottie](<https://devfeed.tech/tags/lottie.md>), [models](<https://devfeed.tech/tags/models.md>), [natural-language](<https://devfeed.tech/tags/natural-language.md>), [rendering](<https://devfeed.tech/tags/rendering.md>), [speculative-decoding](<https://devfeed.tech/tags/speculative-decoding.md>), [svg](<https://devfeed.tech/tags/svg.md>), [tokens](<https://devfeed.tech/tags/tokens.md>), [vector-graphics](<https://devfeed.tech/tags/vector-graphics.md>), [verification](<https://devfeed.tech/tags/verification.md>), [window](<https://devfeed.tech/tags/window.md>)

### AI overview

Sightline speeds lossless SVG generation by pairing verification windows as long as the draft model's horizon with draft trees wide enough to fill them. The approach is motivated by SVG decoding behavior: speculative drafts often pass completely, while errors are rare, clustered, and occur deeper into a draft than in natural language. Across text-to-SVG and image-to-SVG tests with three models on six benchmarks, Sightline was the fastest compared decoder in every setting, including an SArena comparison using shared loop control. It also led in almost every reported setting across six models from four other families generating SVG, compressed SVG tokens, or Lottie JSON. The analysis attributes gains to shorter windows cutting off useful rounds in both loop-free content and loops. The abstract also reports that speculative decoding makes rendering-based sample selection faster than likelihood-based selection.

### Source excerpt

arXiv:2611.23807v1 Announce Type: cross Abstract: Scalable Vector Graphics (SVG) generation models write a graphic as code with one forward pass per token, so drawing is slow. Speculative decoding speeds them up losslessly, and block drafters guess many tokens per pass. On SVG, many of their rounds accept the whole draft and end only at the verification window. We find that their misses are rare and clumped, and that they stay accurate far deeper into a block than on natural language. A run-length identity and three propositions show that a longer window pays once the candidates are wide, up to the drafter's horizon. We propose Sightline, which pairs a window as long as the drafter's horizon with a draft tree wide enough to fill it. Experiments cover text-to-SVG and image-to-SVG with three models on six benchmarks from MMSVG, MMSVG-2M and SArena. In every setting, Sightline is the fastest of the compared lossless decoders. On SArena, it stays the fastest when every decoder shares one loop control. It also leads in almost every setting of six models from four other families, which write SVG code, compressed SVG tokens or Lottie JSON. Analyses trace the lead to rounds that shorter windows cut off, both in loop-free content and in loops. Speculative decoding also saves time for choosing samples by rendering, which beats choosing by likelihood. These results suggest that speculative decoding on vector graphics should verify as far as the drafter sees. Code and data are available at https://github.com/LiZaiyuan0619/Sightline.

## Java4Shell: when base64 decoding leads to remote code execution?

DevFeed: [Java4Shell: when base64 decoding leads to remote code execution?](<https://devfeed.tech/articles/java4shell-when-base64-decoding-leads-to-remote-code-execution-144011.md>)

Original publisher: [Read original article](<https://lemire.me/blog/2026/11/18/java4shell-when-base64-decoding-leads-to-remote-code-execution/>)

Author: Daniel Lemire

Published: 2026-11-18T19:43:03Z

Content type: article

Language: en

Sources: [Daniel Lemire](<https://devfeed.tech/sources/daniel-lemire.md>)

Topics: [openjdk](<https://devfeed.tech/topics/openjdk.md>), [Decoding](<https://devfeed.tech/topics/decoding.md>), [code analysis](<https://devfeed.tech/topics/code-analysis.md>), [Exception](<https://devfeed.tech/topics/exception.md>), [Pull Request](<https://devfeed.tech/topics/pull-request.md>)

Tags: [amd-epyc](<https://devfeed.tech/tags/amd-epyc.md>), [assembly](<https://devfeed.tech/tags/assembly.md>), [avx](<https://devfeed.tech/tags/avx.md>), [avx-512-vbmi](<https://devfeed.tech/tags/avx-512-vbmi.md>), [base64](<https://devfeed.tech/tags/base64.md>), [brave](<https://devfeed.tech/tags/brave.md>), [bug](<https://devfeed.tech/tags/bug.md>), [bun](<https://devfeed.tech/tags/bun.md>), [chrome](<https://devfeed.tech/tags/chrome.md>), [chromium](<https://devfeed.tech/tags/chromium.md>), [cve](<https://devfeed.tech/tags/cve.md>), [cve-2026-114938](<https://devfeed.tech/tags/cve-2026-114938.md>), [decoding](<https://devfeed.tech/tags/decoding.md>), [deno](<https://devfeed.tech/tags/deno.md>), [edge](<https://devfeed.tech/tags/edge.md>), [epyc](<https://devfeed.tech/tags/epyc.md>), [fast](<https://devfeed.tech/tags/fast.md>), [gold](<https://devfeed.tech/tags/gold.md>), [openjdk](<https://devfeed.tech/tags/openjdk.md>), [reading](<https://devfeed.tech/tags/reading.md>)

### AI overview

The excerpt describes CVE-2026-114938, an OpenJDK Base64 decoder bug present since JDK 17.0.6: on x86-64 with AVX-512 VBMI, decoding 128 "A" characters followed by "=" allocates 98 bytes, then a malformed tail length produces a 64-byte masked store at offset 96. Two bytes fit; 62 are written past the array before Java throws IllegalArgumentException. The author says claims of a major security vulnerability are unverified. The excerpt also contrasts Java's fast path, which stops at whitespace, with a newer SIMD decoder that compacts whitespace; it lists browser and runtime deployments and reports 30x and 39x speedups for 64 KiB MIME input on two named processors.

### Source excerpt

Let us start with an example that triggers the issue. byte[] src = ("A".repeat(128) + "=").getBytes(StandardCharsets.US_ASCII); Base64.getDecoder().decode(src); One hundred and twenty-eight letters A and one =. The string is not valid base64, and a correct decoder rejects it. OpenJDK writes past the end of the array anyway. The bug is CVE-2026-114938. It has been in ... Continue reading Java4Shell: when base64 decoding leads to remote code execution?

## Remarkable Observatory Receives Orbital Radio

DevFeed: [Remarkable Observatory Receives Orbital Radio](<https://devfeed.tech/articles/remarkable-observatory-receives-orbital-radio-140942.md>)

Original publisher: [Read original article](<https://hackaday.com/2026/11/18/remarkable-observatory-receives-orbital-radio/>)

Author: Ryan Flowers

Published: 2026-11-18T16:30:34Z

Content type: news

Language: en

Sources: [Hackaday](<https://devfeed.tech/sources/hackaday.md>)

Topics: [communications](<https://devfeed.tech/topics/communications.md>), [Decoding](<https://devfeed.tech/topics/decoding.md>), [recording](<https://devfeed.tech/topics/recording.md>), [Concurrency](<https://devfeed.tech/topics/concurrency.md>)

Tags: [antenna](<https://devfeed.tech/tags/antenna.md>), [claude](<https://devfeed.tech/tags/claude.md>), [claude-code](<https://devfeed.tech/tags/claude-code.md>), [esp32](<https://devfeed.tech/tags/esp32.md>), [github](<https://devfeed.tech/tags/github.md>), [hardware](<https://devfeed.tech/tags/hardware.md>), [image](<https://devfeed.tech/tags/image.md>), [iss](<https://devfeed.tech/tags/iss.md>), [jobs](<https://devfeed.tech/tags/jobs.md>), [observatory](<https://devfeed.tech/tags/observatory.md>), [open-source](<https://devfeed.tech/tags/open-source.md>), [python](<https://devfeed.tech/tags/python.md>), [radio](<https://devfeed.tech/tags/radio.md>), [radio-hacks](<https://devfeed.tech/tags/radio-hacks.md>), [receiver](<https://devfeed.tech/tags/receiver.md>), [reciever](<https://devfeed.tech/tags/reciever.md>), [rf](<https://devfeed.tech/tags/rf.md>), [rtl-sdr](<https://devfeed.tech/tags/rtl-sdr.md>), [satellite](<https://devfeed.tech/tags/satellite.md>), [satellites](<https://devfeed.tech/tags/satellites.md>), [sdr](<https://devfeed.tech/tags/sdr.md>), [software](<https://devfeed.tech/tags/software.md>), [spectrum](<https://devfeed.tech/tags/spectrum.md>)

### AI overview

Starlab Observatory automates satellite and radio monitoring with software-defined radios, an ESP32-controlled antenna switch, and a dedicated Ubuntu computer. Its software predicts satellite passes, selects the receiver and antenna combination, and logs and processes received data; multiple jobs can run at once. The project grew from an attempt to receive ISS image transmissions with a cheap RTL-SDR, which had mixed results. The excerpt identifies it as an open source GitHub project, but does not describe its full capabilities or provide repository details.

### Source excerpt

There's nothing quite so nostalgic as the time honored tradition of tuning across the HF bands, picking out signals from the noise. Add satellites and VHF/UHF to the mix, then ...read more

## MegNIST, A Benchmark for Non-Invasive Inner Speech Decoding

DevFeed: [MegNIST, A Benchmark for Non-Invasive Inner Speech Decoding](<https://devfeed.tech/articles/megnist-a-benchmark-for-non-invasive-inner-speech-decoding-139164.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.21399>)

Author: Teyun Kwon, Pratik Somaiya, Gereon Elvers, Francesco Mantegna, Anna Camera, Sven Braeutigam, Mark Woolrich, Oiwi Parker Jones

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

Content type: article

Language: en

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

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Decoding](<https://devfeed.tech/topics/decoding.md>), [PyTorch](<https://devfeed.tech/topics/pytorch.md>), [reproducibility](<https://devfeed.tech/topics/reproducibility.md>)

Tags: [benchmark](<https://devfeed.tech/tags/benchmark.md>), [benchmark-dataset](<https://devfeed.tech/tags/benchmark-dataset.md>), [classification](<https://devfeed.tech/tags/classification.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [dataset](<https://devfeed.tech/tags/dataset.md>), [decoding](<https://devfeed.tech/tags/decoding.md>), [format](<https://devfeed.tech/tags/format.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [python](<https://devfeed.tech/tags/python.md>), [pytorch](<https://devfeed.tech/tags/pytorch.md>), [recordings](<https://devfeed.tech/tags/recordings.md>), [speech](<https://devfeed.tech/tags/speech.md>), [tensors](<https://devfeed.tech/tags/tensors.md>)

### AI overview

MegNIST provides 12,000 MEG trials from one participant imagining the digits zero through nine across four sessions, offering a benchmark for non-invasive inner-speech decoding. Recordings have 306 channels sampled at 250 Hz and are split by experimental run into 10,000 training, 1,000 validation, and 1,000 test trials to preserve temporal independence. The data are available in BIDS and serialized-tensor formats; the description also notes example code and a PyTorch-compatible loader in pnpl. Baseline classifiers decode imagined digits significantly above chance. The evidence describes a single participant and digit task, so it does not establish performance across people or broader inner speech.

### Source excerpt

arXiv:2611.21399v1 Announce Type: new Abstract: MegNIST is a benchmark dataset for developing and evaluating machine learning methods for inner speech decoding from magnetoencephalography (MEG). The dataset comprises 12,000 trials of a single participant imagining speaking the digits zero through nine, recorded across four sessions. Each trial contains 306-channel MEG recordings sampled at 250 Hz, preprocessed using signal space separation, filtering, and downsampling. Data are provided as pre-defined training (10,000 trials), validation (1,000 trials), and test (1,000 trials) splits, with entire experimental runs assigned to each split to preserve temporal independence. The dataset is available in both BIDS format for neuroimaging analysis and as serialised tensors for machine learning workflows. Baseline classification results demonstrate that imagined digits can be decoded significantly above chance. Standardised splits, validated baselines, example code, and a PyTorch-compatible loader in the pnpl Python library enable straightforward integration with machine learning frameworks, lowering barriers for researchers to rapidly develop and benchmark inner speech decoding methods for non-invasive brain-computer interfaces.

## Inference Equivalence and Quotient Coding

DevFeed: [Inference Equivalence and Quotient Coding](<https://devfeed.tech/articles/inference-equivalence-and-quotient-coding-137657.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.21096>)

Author: Zibo Meng, Andrew C. M. Austin, Alin Achim

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

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: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Decoding](<https://devfeed.tech/topics/decoding.md>)

Tags: [bank](<https://devfeed.tech/tags/bank.md>), [canonical](<https://devfeed.tech/tags/canonical.md>), [compatibility](<https://devfeed.tech/tags/compatibility.md>), [compression](<https://devfeed.tech/tags/compression.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [decoding](<https://devfeed.tech/tags/decoding.md>), [distribution](<https://devfeed.tech/tags/distribution.md>), [frontier](<https://devfeed.tech/tags/frontier.md>), [images](<https://devfeed.tech/tags/images.md>), [inference](<https://devfeed.tech/tags/inference.md>), [learning](<https://devfeed.tech/tags/learning.md>), [mechanism](<https://devfeed.tech/tags/mechanism.md>), [partition](<https://devfeed.tech/tags/partition.md>), [precision](<https://devfeed.tech/tags/precision.md>)

### AI overview

The paper defines which distinctions between sets of interpretations compatible with an observation must be retained to preserve every valid task claim. Under a closedness condition, inference validity is characterized by distance to realizable falsifying measurements; for finite claims ordered by specificity, an Exact Inference Frontier generates all valid claims. Its canonical quotient code groups inference-equivalent compatibility sets into the coarsest partition that permits exact decoding, minimizing distinguishable code values rather than storage or computation. The authors also analyze errors from learned compatibility sets and an asymmetric learning surrogate under distribution shifts. In reported evaluations, 145 sets map to 31 quotient codes with unchanged valid claims; on HRSID, 1,913 candidate sets produce 36 signatures over seven claims, while visual 3-NN reaches 87.31% precision on issued claims.

### Source excerpt

arXiv:2611.21096v1 Announce Type: new Abstract: Incomplete observations can admit multiple physical interpretations without making every task conclusion ambiguous. Recovering all compatible scenes is often unnecessary, whereas indiscriminate compression may change which claims are justified. We ask which distinctions among observation-compatible interpretation sets must be retained to preserve every valid conclusion of a specified task. We first define inference validity as truth across all physically compatible states and characterize it, under a closedness condition, by distances to physically realizable falsifying measurements. For a finite semantic claim family ordered by specificity, the Exact Inference Frontier forms the unique minimal antichain generating all valid claims. We then define inference equivalence between compatibility sets and construct a canonical quotient code from their frontiers. Among deterministic codes admitting exact inference decoding, it induces the coarsest partition, minimizing the number of distinguishable code values rather than storage or computation. Finally, we characterize inference errors from learned compatibility sets, derive conditions for exact inference preservation despite set errors, and evaluate an asymmetric learning surrogate under distribution shifts. On a registered finite bank, 145 compatibility sets yield 31 quotient codes with unchanged valid-claim sets. Across geometry, mechanism, and joint shifts, learned estimates exhibit different accuracy-specificity trade-offs. On 1,962 official HRSID test images, 1,913 distinct candidate sets induce only 36 seven-claim inference signatures, while a visual 3-NN baseline achieves 87.31% precision on issued claims.

## Fast at the Frontier, Wrong at the Fallback: A Commit-Level Analysis of Threshold Decoding in Diffusion Language Models

DevFeed: [Fast at the Frontier, Wrong at the Fallback: A Commit-Level Analysis of Threshold Decoding in Diffusion Language Models](<https://devfeed.tech/articles/fast-at-the-frontier-wrong-at-the-fallback-a-commit-level-analysis-of-threshold-decoding-in-diffusion-language-models-139085.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.20693>)

Author: Roman Makarov, Zhe Xu, Jian Kang

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

Content type: article

Language: en

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

Topics: [Language models](<https://devfeed.tech/topics/language-models.md>), [Decoding](<https://devfeed.tech/topics/decoding.md>)

Tags: [analysis](<https://devfeed.tech/tags/analysis.md>), [code](<https://devfeed.tech/tags/code.md>), [commit](<https://devfeed.tech/tags/commit.md>), [commits](<https://devfeed.tech/tags/commits.md>), [confidence](<https://devfeed.tech/tags/confidence.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [decoder](<https://devfeed.tech/tags/decoder.md>), [decoding](<https://devfeed.tech/tags/decoding.md>), [diffusion](<https://devfeed.tech/tags/diffusion.md>), [errors](<https://devfeed.tech/tags/errors.md>), [failure](<https://devfeed.tech/tags/failure.md>), [failures](<https://devfeed.tech/tags/failures.md>), [fallback](<https://devfeed.tech/tags/fallback.md>), [fast](<https://devfeed.tech/tags/fast.md>), [frontier](<https://devfeed.tech/tags/frontier.md>), [language](<https://devfeed.tech/tags/language.md>), [language-models](<https://devfeed.tech/tags/language-models.md>), [math](<https://devfeed.tech/tags/math.md>), [model](<https://devfeed.tech/tags/model.md>), [models](<https://devfeed.tech/tags/models.md>), [position](<https://devfeed.tech/tags/position.md>)

### AI overview

The study traces token commits from four off-the-shelf diffusion language models on math, code, translation, and summarization tasks. Most tokens become confident near the leftmost masked position, so restricting commits to that frontier still allows several tokens per step and is reported as about as fast as committing anywhere. Accuracy loss is linked to committing tokens before the text to their left; committing one token per step does not remove it. Fallback commits account for 11-35% of commits but initiate 61-72% of failures avoided by left-to-right decoding, with early end-of-sequence commits especially harmful. Changing the step where a fallback-triggered failure begins repairs answers more often than changing a similar earlier step. Frontier-only commits recover most of the lost accuracy across all four models.

### Source excerpt

arXiv:2611.20693v1 Announce Type: cross Abstract: Diffusion language models (DLMs) can write several tokens per forward pass. Open-weight DLMs commit, in any order, every masked position whose confidence passes a threshold. If none does, a fallback commits the most confident one, and the commits are never revised. Wide windows let the decoder commit far ahead of the text, which should speed up decoding but often costs accuracy. It is unclear where the model is confident and which commits cause errors. We trace every commit of four off-the-shelf DLMs on math, code, translation and summarization, and rerun failing answers with one step changed. Most tokens become confident only near the frontier, the leftmost masked position, so committing only there, still several tokens per step, is about as fast as committing anywhere. The accuracy loss comes from committing tokens before the text to their left, and committing one token per step does not reduce it. The worst case is an early end-of-sequence token that cuts the answer short. Fallback commits are 11-35% of commits but start 61-72% of the failures that left-to-right decoding avoids. Changing only the step where such a failure starts repairs it more often than changing a similar earlier step. Committing only at the frontier recovers most of the lost accuracy on all four models. Our code is available at https://github.com/rmakarovv/fast-at-the-frontier.

## LZn : Robust LoRa Frame Synchronization Under Frame Collisions and Ultra-Low SNR Conditions

DevFeed: [LZn : Robust LoRa Frame Synchronization Under Frame Collisions and Ultra-Low SNR Conditions](<https://devfeed.tech/articles/lzn-robust-lora-frame-synchronization-under-frame-collisions-and-ultra-low-snr-conditions-137013.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2604.27672>)

Author: Jos\\'e \\'Alamos, Thomas C. Schmidt, Matthias W\\"ahlisch

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

Content type: article

Language: en

Sources: [arXiv: Networking](<https://devfeed.tech/sources/arxiv-networking.md>)

Topics: [synchronization](<https://devfeed.tech/topics/synchronization.md>), [Decoding](<https://devfeed.tech/topics/decoding.md>), [Simulation](<https://devfeed.tech/topics/simulation.md>)

Tags: [cost](<https://devfeed.tech/tags/cost.md>), [cs-ni](<https://devfeed.tech/tags/cs-ni.md>), [cs-sy](<https://devfeed.tech/tags/cs-sy.md>), [datasets](<https://devfeed.tech/tags/datasets.md>), [decoding](<https://devfeed.tech/tags/decoding.md>), [deployments](<https://devfeed.tech/tags/deployments.md>), [detection](<https://devfeed.tech/tags/detection.md>), [eess-sy](<https://devfeed.tech/tags/eess-sy.md>), [frame](<https://devfeed.tech/tags/frame.md>), [intersection](<https://devfeed.tech/tags/intersection.md>), [lora](<https://devfeed.tech/tags/lora.md>), [probability](<https://devfeed.tech/tags/probability.md>), [range](<https://devfeed.tech/tags/range.md>), [real-time](<https://devfeed.tech/tags/real-time.md>), [real-world](<https://devfeed.tech/tags/real-world.md>), [recovery](<https://devfeed.tech/tags/recovery.md>), [robust](<https://devfeed.tech/tags/robust.md>), [scheme](<https://devfeed.tech/tags/scheme.md>), [signal](<https://devfeed.tech/tags/signal.md>), [simulations](<https://devfeed.tech/tags/simulations.md>), [snr](<https://devfeed.tech/tags/snr.md>), [symbols](<https://devfeed.tech/tags/symbols.md>), [synchronization](<https://devfeed.tech/tags/synchronization.md>)

### AI overview

LZn targets LoRa frame synchronization when same-spreading-factor packets collide or arrive at very low SNR. It uses a minimum-based spectral intersection across consecutive symbols. Preamble search adds modest overhead, while fine synchronization costs two to three orders of magnitude less than in existing collision-tolerant receivers. In simulations and three real-world datasets, LZn improves detection sensitivity by up to 10 dB and detection probability by up to 1.54x. Against TnB, the next-best collision-tolerant scheme, it improves real-world decoding by 3.46x in the most challenging single-user scenario and up to 1.22x in collision scenarios. The abstract reports compatibility with real-time requirements.

### Source excerpt

arXiv:2604.27672v2 Announce Type: replace Abstract: LoRa has become a widely adopted wireless modulation scheme in LPWANs due to its low cost, long range, and minimal transmission power. However, collisions between frames of the same spreading factor -- common in dense LoRa deployments -- prevent conventional LoRa receivers from detecting and correctly decoding frames. Recent work has introduced methods to improve recovery, yet their detection stage degrades sharply under low signal-to-noise ratio (SNR) and high collision rates. In this work, we introduce LZn, a synchronization scheme driven by a minimum-based spectral intersection across consecutive symbols. While preamble search incurs a modest overhead, LZn reduces the cost of fine synchronization by two to three orders of magnitude compared to existing collision-tolerant receivers. Our method enables robust frame synchronization even under multiple packet overlaps or extremely low SNR conditions. We evaluate LZn on simulations and three independent, real-world LoRa datasets. LZn improves detection sensitivity by up to 10 dB and increases detection probability by up to 1.54x. In real-world datasets, LZn improves decoding by 3.46x in the most challenging single-user scenario and up to 1.22x in collision scenarios compared to the second best collision-tolerant scheme (TnB). These results demonstrate that LZn substantially improves the frame recovery of LoRa receivers, while remaining compatible with real-time requirements.

## SpecRelay: Amortize or Avoid the Round Trip in Edge-Cloud Speculative Decoding

DevFeed: [SpecRelay: Amortize or Avoid the Round Trip in Edge-Cloud Speculative Decoding](<https://devfeed.tech/articles/specrelay-amortize-or-avoid-the-round-trip-in-edge-cloud-speculative-decoding-137368.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.18963>)

Author: Jai Jain, Chihyo Ahn, Yong Liu, Charles Chen, Duc N. M Hoang, Ajay Kumar Jaiswal, Fatih Ilhan, Han-Byul Kim, Lauren Hannah, Richa Dixit, Minsik Cho, Yingyan Celine Lin, Hyesoon Kim

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

Content type: news

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: [Model Optimizer](<https://devfeed.tech/topics/model-optimizer.md>), [Local AI](<https://devfeed.tech/topics/local-ai.md>), [On-device AI](<https://devfeed.tech/topics/on-device-ai.md>), [Cloud](<https://devfeed.tech/topics/cloud.md>), [Decoding](<https://devfeed.tech/topics/decoding.md>)

Tags: [asynchronous](<https://devfeed.tech/tags/asynchronous.md>), [autoregressive](<https://devfeed.tech/tags/autoregressive.md>), [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [cloud](<https://devfeed.tech/tags/cloud.md>), [completion](<https://devfeed.tech/tags/completion.md>), [concurrently](<https://devfeed.tech/tags/concurrently.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [decoding](<https://devfeed.tech/tags/decoding.md>), [device](<https://devfeed.tech/tags/device.md>), [draft](<https://devfeed.tech/tags/draft.md>), [edge](<https://devfeed.tech/tags/edge.md>), [edge-cloud](<https://devfeed.tech/tags/edge-cloud.md>), [full](<https://devfeed.tech/tags/full.md>), [generation](<https://devfeed.tech/tags/generation.md>), [inference](<https://devfeed.tech/tags/inference.md>), [latency](<https://devfeed.tech/tags/latency.md>), [llm](<https://devfeed.tech/tags/llm.md>), [llm-inference](<https://devfeed.tech/tags/llm-inference.md>), [local](<https://devfeed.tech/tags/local.md>), [low-latency](<https://devfeed.tech/tags/low-latency.md>), [speculative-decoding](<https://devfeed.tech/tags/speculative-decoding.md>)

### AI overview

SpecRelay tackles round-trip latency in edge-cloud speculative decoding in two ways. For cloud-bound queries, its asynchronous pipeline keeps fixed-budget draft trees in flight while prior verifications are pending; the cloud discards stale requests before running the target model. This schedule alone raises throughput by up to 50% over a reproduced synchronous baseline across multiple benchmarks. An RTT-conditioned Latency-Aware Router also chooses among on-device direct generation, Chain-of-Draft, full reasoning, and cloud speculative decoding under a configurable accuracy-latency objective. It runs alongside draft-model prefill, adding no inference latency. At matched target-model accuracy with 250 ms of added round-trip latency, the full system reaches 1.95x baseline throughput on SPROUT and 1.55x on OlympiadBench, sending 47% and 39% of queries to the cloud, respectively.

### Source excerpt

arXiv:2611.18963v1 Announce Type: new Abstract: Edge-cloud speculative decoding moves autoregressive drafting to an edge device while a larger cloud model verifies draft tokens in parallel, so that edge drafting and cloud verification overlap under a synchronous draft-verify schedule. Beyond low-latency local networks, however, the next verification request of that schedule can be sent only after the previous reply arrives, so every round still waits out a full round trip, and throughput falls as the network RTT grows. To recover this loss, we present SpecRelay, an edge-cloud LLM inference system that addresses this cost at two levels: it amortizes round trips when cloud verification is necessary and avoids them when local reasoning suffices. For cloud-bound queries, an asynchronous verification pipeline keeps fixed-budget draft trees in flight while earlier replies are still outstanding, and the cloud rejects stale requests before the target-model forward pass (the asynchronous schedule). At the query level, an RTT-conditioned router (the Latency-Aware Router) estimates quality and completion latency for three on-device modes (direct generation, Chain-of-Draft, and full reasoning) and cloud speculative decoding, selecting among them under a configurable accuracy-latency objective without adding latency during inference, since the router runs concurrently with the draft-model prefill. The asynchronous schedule alone improves decoding throughput by up to 50% over a reproduced synchronous baseline across multiple benchmarks. At matched target-model accuracy and 250 ms of added round-trip latency, the full system with the router delivers 1.95x and 1.55x the throughput of the same baseline on SPROUT and OlympiadBench, respectively, while sending only 47% and 39% of the queries to the cloud.

## ForeSay: Tell the Future in Words, Branch the Details in Noise for Precipitation Nowcasting

DevFeed: [ForeSay: Tell the Future in Words, Branch the Details in Noise for Precipitation Nowcasting](<https://devfeed.tech/articles/foresay-tell-the-future-in-words-branch-the-details-in-noise-for-precipitation-nowcasting-137616.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.20815>)

Author: Xudong Ling, Wenjie Luo, Chaorong Li, Tianxi Huang, Tao He, Guiduo Duan

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

Content type: article

Language: en

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

Topics: [Predictive AI](<https://devfeed.tech/topics/predictive-ai.md>), [AI Chat](<https://devfeed.tech/topics/ai-chat.md>), [Decoding](<https://devfeed.tech/topics/decoding.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [best](<https://devfeed.tech/tags/best.md>), [branch](<https://devfeed.tech/tags/branch.md>), [condition](<https://devfeed.tech/tags/condition.md>), [consensus](<https://devfeed.tech/tags/consensus.md>), [cost](<https://devfeed.tech/tags/cost.md>), [coupling](<https://devfeed.tech/tags/coupling.md>), [critical](<https://devfeed.tech/tags/critical.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [csi](<https://devfeed.tech/tags/csi.md>), [datasets](<https://devfeed.tech/tags/datasets.md>), [decoding](<https://devfeed.tech/tags/decoding.md>), [details](<https://devfeed.tech/tags/details.md>), [diversity](<https://devfeed.tech/tags/diversity.md>), [evolution](<https://devfeed.tech/tags/evolution.md>), [explore](<https://devfeed.tech/tags/explore.md>), [fine](<https://devfeed.tech/tags/fine.md>), [flow](<https://devfeed.tech/tags/flow.md>), [future](<https://devfeed.tech/tags/future.md>), [noise](<https://devfeed.tech/tags/noise.md>)

### AI overview

ForeSay separates storm-scale evolution from local convective uncertainty in precipitation nowcasts. A vision-language model turns historical radar frames into a textual storm scenario that conditions a Rectified Flow generator. Ensemble members share early sampling steps to establish a common storm trajectory, then branch with fresh noise in later steps for local detail. The method uses self-anchored support-consensus voting instead of pixel averaging to preserve sharp heavy-rain cores. Across six datasets, it reports the best CSI at the highest precipitation thresholds, improving over the strongest baseline by more than 10% to 84%, depending on the dataset. An eight-member ensemble reportedly retains the skill and diversity of eight independent runs while cutting compute from 8x to 5.2x a single forecast.

### Source excerpt

arXiv:2611.20815v1 Announce Type: new Abstract: Precipitation nowcasting involves two distinct scales of uncertainty: macroscopic storm evolution and fine-scale local convective variations. Current generative ensemble methods entangle both scales within independent random noise draws. This coupling causes two practical limitations: every ensemble member must execute the full sampling trajectory from scratch, so an N-member ensemble costs N times a single forecast; moreover, divergent trajectories spatially misalign heavy-rain echoes across members, causing standard pixel-wise averaging to blur critical high-intensity extremes. We present ForeSay, a generative nowcasting framework that decouples macroscopic evolution prediction from local stochastic detail generation. Given historical radar frames, a vision-language model first predicts a textual scenario of future storm dynamics to condition a Rectified Flow generator. Guided by this shared macroscopic prior, ensemble members share early trunk sampling steps to establish the storm trajectory and branch with fresh noise only in later steps to explore local convective details. Because members then differ mainly in local details, self-anchored support-consensus decoding replaces pixel averaging with discrete support voting to preserve sharp heavy-rain cores. Across six datasets, ForeSay achieves the best CSI at the highest precipitation thresholds, with relative improvements over the strongest baseline on each dataset ranging from over 10% to 84%. Our 8-member ensemble retains the forecasting skill and diversity of eight independently sampled runs while reducing computational cost from 8x to 5.2x that of a single forecast.

## IBM, AMD advance quantum error correction decoding at scale

DevFeed: [IBM, AMD advance quantum error correction decoding at scale](<https://devfeed.tech/articles/ibm-amd-advance-quantum-error-correction-decoding-at-scale-136373.md>)

Original publisher: [Read original article](<https://research.ibm.com/blog/pelagos>)

Author: Blake Johnson; Robert Davis

Published: 2026-11-16T15:59:00Z

Content type: news

Language: en

Sources: [IBM Research](<https://devfeed.tech/sources/ibm-research.md>)

Topics: [Quantum Computing](<https://devfeed.tech/topics/quantum-computing.md>), [Decoding](<https://devfeed.tech/topics/decoding.md>), [rocm](<https://devfeed.tech/topics/rocm.md>), [ibm](<https://devfeed.tech/topics/ibm.md>)

Tags: [acceleration](<https://devfeed.tech/tags/acceleration.md>), [accelerators](<https://devfeed.tech/tags/accelerators.md>), [algorithms](<https://devfeed.tech/tags/algorithms.md>), [amd](<https://devfeed.tech/tags/amd.md>), [decoding](<https://devfeed.tech/tags/decoding.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [ibm](<https://devfeed.tech/tags/ibm.md>), [open-source](<https://devfeed.tech/tags/open-source.md>), [quantum](<https://devfeed.tech/tags/quantum.md>), [quantum-error-correction](<https://devfeed.tech/tags/quantum-error-correction.md>), [quantum-research](<https://devfeed.tech/tags/quantum-research.md>), [quantum-software](<https://devfeed.tech/tags/quantum-software.md>), [release](<https://devfeed.tech/tags/release.md>), [scale](<https://devfeed.tech/tags/scale.md>)

### AI overview

IBM and AMD released Pelagos, an open-source framework for evaluating quantum error-correction decoders at scale. Built with Triton, it supports AMD and NVIDIA GPUs and combines GPU acceleration with a reusable software architecture. Multi-GPU work uses process-level orchestration, parallel host-side operations and shared memory to reduce coordination overhead. Tests on AMD Instinct MI300X and MI355X showed up to roughly 7.5x speedup in the GPU decoding stage across eight GPUs versus one; the team also deployed Pelagos on ORNL's Lux HPC system. The excerpt says earlier studies found up to 97x higher throughput than a single CPU core. An accompanying survey reviews decoding methods and platforms including CPUs, GPUs, FPGAs and specialized accelerators, and argues that classical infrastructure and hardware-software co-design are key parts of building fault-tolerant quantum systems.

### Source excerpt

New open-source tool from IBM and AMD helps researchers evaluate quantum error-correction decoders at scale for future fault-tolerant quantum systems.

## Neural Classifiers as Error-Correcting Codes

DevFeed: [Neural Classifiers as Error-Correcting Codes](<https://devfeed.tech/articles/neural-classifiers-as-error-correcting-codes-135882.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.18239>)

Author: Sergey Salishev, Anton Makarov, Oleg Granichin, Renata Avros, Zeev Volkovich

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

Content type: article

Language: en

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

Topics: [Neural Network](<https://devfeed.tech/topics/neural-network.md>), [Deep learning](<https://devfeed.tech/topics/deep-learning.md>), [Deep neural networks](<https://devfeed.tech/topics/deep-neural-networks.md>), [Decoding](<https://devfeed.tech/topics/decoding.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [quantization](<https://devfeed.tech/topics/quantization.md>), [Post-training optimization](<https://devfeed.tech/topics/post-training-optimization.md>)

Tags: [classification](<https://devfeed.tech/tags/classification.md>), [classifiers](<https://devfeed.tech/tags/classifiers.md>), [cs-it](<https://devfeed.tech/tags/cs-it.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [decoding](<https://devfeed.tech/tags/decoding.md>), [deep](<https://devfeed.tech/tags/deep.md>), [distillation](<https://devfeed.tech/tags/distillation.md>), [error-correcting-codes](<https://devfeed.tech/tags/error-correcting-codes.md>), [feature](<https://devfeed.tech/tags/feature.md>), [hamming](<https://devfeed.tech/tags/hamming.md>), [knowledge-distillation](<https://devfeed.tech/tags/knowledge-distillation.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [math-it](<https://devfeed.tech/tags/math-it.md>), [models](<https://devfeed.tech/tags/models.md>), [noise](<https://devfeed.tech/tags/noise.md>), [objectives](<https://devfeed.tech/tags/objectives.md>), [optimization](<https://devfeed.tech/tags/optimization.md>), [quantization](<https://devfeed.tech/tags/quantization.md>), [redundancy](<https://devfeed.tech/tags/redundancy.md>), [shannon](<https://devfeed.tech/tags/shannon.md>), [space](<https://devfeed.tech/tags/space.md>)

### AI overview

The abstract proposes viewing supervised deep classification as error-correcting-code decoding: cross-entropy aligns softmax classifiers with maximum-likelihood decoding, with logits corresponding to Hamming distances to class-specific codewords. It says this framing accounts for standard divergence objectives and offers a geometric basis for decision-boundary stability. Neural Collapse and LeJEPA geometries are described as optimal decoding constellations under isotropic noise, while deep networks learn feature-space codebooks implicitly rather than using classical Error-Correcting Output Codes. The authors propose applying this view to lossless quantization, knowledge distillation, and architectural regularization through code redundancy, and extend it to self-supervised encoders. These are claims and proposed applications in the abstract; supporting details are not included.

### Source excerpt

arXiv:2611.18239v1 Announce Type: new Abstract: A unified framework is proposed that characterizes supervised deep classification as the decoding of error-correcting codes in the Shannon sense. Cross-entropy optimization is shown to align softmax classifiers with maximum-likelihood decoding, where logits map to Hamming distances to class-specific codewords. This correspondence explains the emergence of standard divergence objectives (e.g., Kullback-Leibler, Jeffreys) and provides a geometric basis for the stability of decision boundaries. Furthermore, the geometries of Neural Collapse and LeJEPA are identified as optimal constellations for decoding under isotropic noise. Unlike classical Error-Correcting Output Codes, deep networks are shown to learn effective codebooks in the feature space implicitly. This insight is leveraged to derive novel criteria for lossless quantization, improved knowledge distillation, and architectural regularization based on code redundancy. Finally, this perspective is extended to self-supervised encoders, offering a cohesive explanation for the geometry of latent representations in modern neural models.

## PACT: A Decoder-Confidence-Guided Two-Round Receiver for Joint Channel and Phase-Noise Estimation in Sub-THz OFDM

DevFeed: [PACT: A Decoder-Confidence-Guided Two-Round Receiver for Joint Channel and Phase-Noise Estimation in Sub-THz OFDM](<https://devfeed.tech/articles/pact-a-decoder-confidence-guided-two-round-receiver-for-joint-channel-and-phase-noise-estimation-in-sub-thz-ofdm-135902.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.18450>)

Author: Ruilin Wang, Xiaodai Dong

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

Content type: article

Language: en

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

Topics: [Transformer](<https://devfeed.tech/topics/transformer.md>), [Neural Network](<https://devfeed.tech/topics/neural-network.md>), [Decoding](<https://devfeed.tech/topics/decoding.md>), [fourier analysis](<https://devfeed.tech/topics/fourier-analysis.md>)

Tags: [attention](<https://devfeed.tech/tags/attention.md>), [bandwidth](<https://devfeed.tech/tags/bandwidth.md>), [channel](<https://devfeed.tech/tags/channel.md>), [confidence](<https://devfeed.tech/tags/confidence.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [decoder](<https://devfeed.tech/tags/decoder.md>), [eess-sp](<https://devfeed.tech/tags/eess-sp.md>), [estimator](<https://devfeed.tech/tags/estimator.md>), [fft](<https://devfeed.tech/tags/fft.md>), [learning](<https://devfeed.tech/tags/learning.md>), [linear](<https://devfeed.tech/tags/linear.md>), [neural](<https://devfeed.tech/tags/neural.md>), [noise](<https://devfeed.tech/tags/noise.md>), [phase-noise](<https://devfeed.tech/tags/phase-noise.md>), [pilot](<https://devfeed.tech/tags/pilot.md>), [receiver](<https://devfeed.tech/tags/receiver.md>), [selective-fading](<https://devfeed.tech/tags/selective-fading.md>), [settings](<https://devfeed.tech/tags/settings.md>), [tests](<https://devfeed.tech/tags/tests.md>), [time](<https://devfeed.tech/tags/time.md>), [transformer](<https://devfeed.tech/tags/transformer.md>), [weight](<https://devfeed.tech/tags/weight.md>)

### AI overview

PACT combines a pilot-aware complex transformer that jointly estimates frequency-domain channel and effective time-domain phase noise with banded LMMSE equalization and neural normalized min-sum decoding. Decoder-derived pseudo-pilots and constellation-aware confidence guide a second refinement round; parameters are shared across rounds. At 30 dB Eb/N0, the receiver reports about 70% and 57% lower code-block error rates than the strongest learning-based baseline in two benchmark settings. Frozen-weight tests indicate applicability across tested channel and phase-noise models, carrier frequencies, and FFT/bandwidth configurations.

### Source excerpt

arXiv:2611.18450v1 Announce Type: new Abstract: Frequency-selective fading and oscillator phase noise (PN) jointly impair coded sub-THz OFDM reception. We propose a decoder-confidence-guided two-round receiver combining a {pilot-aware complex transformer} (PACT) estimator, banded linear minimum mean-square error equalization, and neural normalized min-sum decoding. PACT jointly estimates the frequency-domain channel and effective time-domain PN through interacting complex-valued branches. Decoder-derived pseudo-pilots and continuous constellation-aware confidence guide second-round refinement through PACT inputs and attention, with parameters shared across rounds. At $E_b/N_0=30$~dB, the PACT receiver achieves approximately $70\%$ and $57\%$ lower code-block error rates than the best-performing learning-based baseline in the two benchmark settings, respectively. Frozen-weight tests further demonstrate applicability across channel/PN models, carrier frequencies, and FFT/bandwidth configurations.

## Typing Matters: Muscle, Language and Typing Behaviour in Live Keystroke Decoding from sEMG

DevFeed: [Typing Matters: Muscle, Language and Typing Behaviour in Live Keystroke Decoding from sEMG](<https://devfeed.tech/articles/typing-matters-muscle-language-and-typing-behaviour-in-live-keystroke-decoding-from-semg-135202.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.18464>)

Author: Abhilash Durgam, Nyle Siddiqui, Mubarak Shah, Qiushi Fu

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

Content type: article

Language: en

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

Topics: [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Language Engineering](<https://devfeed.tech/topics/language-engineering.md>), [Decoding](<https://devfeed.tech/topics/decoding.md>)

Tags: [autoregressive](<https://devfeed.tech/tags/autoregressive.md>), [behaviour](<https://devfeed.tech/tags/behaviour.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [context](<https://devfeed.tech/tags/context.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [decoder](<https://devfeed.tech/tags/decoder.md>), [decoding](<https://devfeed.tech/tags/decoding.md>), [error](<https://devfeed.tech/tags/error.md>), [fixed](<https://devfeed.tech/tags/fixed.md>), [future](<https://devfeed.tech/tags/future.md>), [improvement](<https://devfeed.tech/tags/improvement.md>), [inference](<https://devfeed.tech/tags/inference.md>), [keys](<https://devfeed.tech/tags/keys.md>), [language](<https://devfeed.tech/tags/language.md>), [language-model](<https://devfeed.tech/tags/language-model.md>), [live](<https://devfeed.tech/tags/live.md>), [model](<https://devfeed.tech/tags/model.md>), [moving](<https://devfeed.tech/tags/moving.md>), [session](<https://devfeed.tech/tags/session.md>), [signal](<https://devfeed.tech/tags/signal.md>), [typing](<https://devfeed.tech/tags/typing.md>)

### AI overview

The proposed sEMG typing decoder predicts keys live with no future context, carrying session history forward instead of processing isolated fixed windows. A state space model combines muscle signals with an autoregressive language model conditioned on text already typed, including typing behaviors such as typos, backspaces, and line breaks. To train per-character predictions without sEMG-to-character transcripts, the authors retokenize WikiText into human-like typing sequences. On emg2qwerty's cross-user benchmark, the model reports a 25.13% character error rate, a 5.4% relative improvement over the previous best with 6.7 times fewer parameters.

### Source excerpt

arXiv:2611.18464v1 Announce Type: new Abstract: Surface electromyography (sEMG) from the wrists can recover typed text without a keyboard. It is usually posed as a signal-inference problem: characters are read out of the muscle signal and character error rate is reported offline. We argue that this framing removes most of the problem. The decoder is constrained simultaneously by the muscles, by the language being typed, and by typing behaviour such as typos, corrections, backspaces and line breaks, and a usable system must predict keys live while the hand keeps moving. Existing systems fall short on two counts: (1) they process sEMG in fixed windows and carry nothing from one window to the next, discarding context accumulated over a session, and (2) they lack any understanding of the linguistic structure that drives typing, so every keystroke is inferred from the muscle signal alone. We propose a fully causal, zero-lookahead sEMG typing decoder built on a state space model (SSM), in which a single model decodes a whole session without windowed partitions, conditioned on what has already been typed. We achieve this by fusing a traditional typing decoder with an efficient autoregressive language model, instilling both sEMG and linguistic understanding in one model. To train such a model for per-character typing rather than whole-word text, we introduce a retokenization method that rewrites ordinary prose as human-like typing, with keystrokes, typos, backspaces and line breaks included, converting WikiText into a large typing corpus where none existed. Our model, having never seen a single sEMG-to-character transcript in typing, achieves a state-of-the-art character error rate of 25.13% on emg2qwerty's cross-user benchmark, a 5.4% relative improvement over the previous best with 6.7 times fewer parameters, while running strictly causally with no future context.

## Toward Interference-Agnostic RF Source Separation

DevFeed: [Toward Interference-Agnostic RF Source Separation](<https://devfeed.tech/articles/toward-interference-agnostic-rf-source-separation-135153.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.17950>)

Author: Maxence Albert, Rapha\\"el Duroselle, Thibaud Ehret, Olivier Boeffard, Jean-Fran\\c{c}ois Bonastre

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

Content type: article

Language: en

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

Topics: [Decoding](<https://devfeed.tech/topics/decoding.md>), [fourier transform](<https://devfeed.tech/topics/fourier-transform.md>), [Resilience](<https://devfeed.tech/topics/resilience.md>)

Tags: [architectures](<https://devfeed.tech/tags/architectures.md>), [band](<https://devfeed.tech/tags/band.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [eess-sp](<https://devfeed.tech/tags/eess-sp.md>), [rf](<https://devfeed.tech/tags/rf.md>), [separator](<https://devfeed.tech/tags/separator.md>), [signal](<https://devfeed.tech/tags/signal.md>), [source](<https://devfeed.tech/tags/source.md>)

### AI overview

Specialized RF signal separators depend on knowing the signal-interference pair, and the study reports that using one on an unseen interference usually fails, often leaving the signal worse than no processing. The authors train a single separator across all pairs without identifying the interference at test time, using both the ICASSP 2024 RF Challenge WaveNet and an adapted band-split transformer; both show similar results. At the size of one specialized model, the unified separator keeps bit error rates close on nearly every tested pair, performing better on one and worse on another. For an interference absent from all training sets, it matches the best specialized model, without requiring a choice of model at deployment.

### Source excerpt

arXiv:2611.17950v1 Announce Type: new Abstract: Separating a radio-frequency signal of interest from a co-channel interference is mostly addressed with specialized separators, each trained for one pair of signal and interference types and therefore relying on a label that a heavily jammed receiver cannot identify. We show that this reliance is not benign: applied to an interference it was not trained on, such a separator usually fails to recover the signal, and in most of the cases we measure it leaves the signal worse than no processing at all. We then train a single separator over every pair, giving it no indication at test time of which one it faces, on two architectures with no component in common: the convolutional WaveNet of the ICASSP 2024 RF Challenge, and a band-split transformer that we adapt from music source separation. Both behave the same way. At the size of one specialized model, the unified separator stays close to the whole models it replaces, keeping the bit error rate close on almost every pair, better on one and worse on another. On an interference absent from every training set, it does as well as the best specialized model, without anyone having to choose which one to deploy.

## Fast-dLLM++: Fr\\'echet Profile Decoding for Faster Diffusion LLM Inference

DevFeed: [Fast-dLLM++: Fr\\'echet Profile Decoding for Faster Diffusion LLM Inference](<https://devfeed.tech/articles/fast-dllm-fr-echet-profile-decoding-for-faster-diffusion-llm-inference-131924.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2606.02955>)

Author: Siva Rajesh Kasa, Yasong Dai, Sumit Negi, Hongdong Li

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

Content type: article

Language: en

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

Topics: [LLM optimization](<https://devfeed.tech/topics/llm-optimization.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [Decoding](<https://devfeed.tech/topics/decoding.md>)

Tags: [cache](<https://devfeed.tech/tags/cache.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>), [decoding](<https://devfeed.tech/tags/decoding.md>), [diffusion](<https://devfeed.tech/tags/diffusion.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [extension](<https://devfeed.tech/tags/extension.md>), [factor](<https://devfeed.tech/tags/factor.md>), [frontier](<https://devfeed.tech/tags/frontier.md>), [generation](<https://devfeed.tech/tags/generation.md>), [humaneval](<https://devfeed.tech/tags/humaneval.md>), [inference](<https://devfeed.tech/tags/inference.md>), [llm](<https://devfeed.tech/tags/llm.md>), [mbpp](<https://devfeed.tech/tags/mbpp.md>), [parallel](<https://devfeed.tech/tags/parallel.md>), [process](<https://devfeed.tech/tags/process.md>), [profile](<https://devfeed.tech/tags/profile.md>), [selection](<https://devfeed.tech/tags/selection.md>), [speed](<https://devfeed.tech/tags/speed.md>), [theoretical](<https://devfeed.tech/tags/theoretical.md>), [throughput](<https://devfeed.tech/tags/throughput.md>), [token](<https://devfeed.tech/tags/token.md>)

### AI overview

Fast-dLLM++ replaces Fast-dLLM's weakest-token confidence rule with Fréchet profile decoding, which selects parallel commit sets using the full sorted confidence profile. The rule matches the original when selected tokens have equal confidence and provides a provable heterogeneity bonus when their confidences differ. This training-free change leaves the model, diffusion process, and cache implementation unchanged. On GSM8K, MATH, HumanEval, and MBPP with LLaDA-8B, the authors report an improved accuracy-throughput frontier and up to 37% higher throughput at comparable accuracy. Evidence is from the supplied abstract.

### Source excerpt

arXiv:2606.02955v3 Announce Type: replace-cross Abstract: Diffusion large language models promise parallel token generation, yet inference remains bottlenecked by deciding which masked tokens can be safely committed together. Fast-dLLM addressed this with KV caching and confidence-guided parallel decoding, but its decoding theory uses a homogeneous high-confidence assumption that effectively reduces each candidate set to its weakest selected token. We argue that this leaves speed on the table because real decoding steps exhibit heterogeneous confidence profiles. We propose Fast-dLLM++, a training-free extension that introduces Fr\'echet profile decoding: selecting parallel commit sets from the full sorted confidence profile rather than a single worst-case confidence. The resulting rule is a heterogeneous-confidence generalization of Fast-dLLM's factor selector and it recovers the previous rule exactly in the equal-confidence case and adds a provable heterogeneity bonus when the selected tokens have uneven confidences. Fast-dLLM++ leaves the model, diffusion process, and cache implementation entirely unchanged, making it a drop-in replacement for existing Fast-dLLM decoding. Experiments on GSM8K, MATH, HumanEval, and MBPP with the LLaDA-8B model show that the theoretical improvement translates directly into empirical gains: profile-aware selection improves the accuracy-throughput frontier by exploiting safe parallelism that weakest-token rules miss, achieving up to 37% higher throughput at comparable accuracy. Our code release is at https://github.com/Ringo-Star/FastdLLM_plusplus.

## Three Lincoln Laboratory technologies receive 2026 R&D 100 Awards

DevFeed: [Three Lincoln Laboratory technologies receive 2026 R&D 100 Awards](<https://devfeed.tech/articles/three-lincoln-laboratory-technologies-receive-2026-r-d-100-awards-131194.md>)

Original publisher: [Read original article](<https://news.mit.edu/2026/lincoln-laboratory-technologies-receive-rd-100-awards-1112>)

Author: Kylie Foy | MIT Lincoln Laboratory

Published: 2026-11-12T21:15:00Z

Content type: news

Language: en

Sources: [MIT AI News](<https://devfeed.tech/sources/mit-ai-news.md>)

Topics: [systems](<https://devfeed.tech/topics/systems.md>), [communications](<https://devfeed.tech/topics/communications.md>), [kasperskyos](<https://devfeed.tech/topics/kasperskyos.md>), [Kernel](<https://devfeed.tech/topics/kernel.md>), [Decoding](<https://devfeed.tech/topics/decoding.md>), [Critical Infrastructure](<https://devfeed.tech/topics/critical-infrastructure.md>), [middleware](<https://devfeed.tech/topics/middleware.md>), [PWA](<https://devfeed.tech/topics/pwa.md>), [Frontier AI](<https://devfeed.tech/topics/frontier-ai.md>)

Tags: [2025](<https://devfeed.tech/tags/2025.md>), [2026](<https://devfeed.tech/tags/2026.md>), [5g](<https://devfeed.tech/tags/5g.md>), [5g-interference](<https://devfeed.tech/tags/5g-interference.md>), [6g](<https://devfeed.tech/tags/6g.md>), [addition](<https://devfeed.tech/tags/addition.md>), [after](<https://devfeed.tech/tags/after.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-engine](<https://devfeed.tech/tags/ai-engine.md>), [air](<https://devfeed.tech/tags/air.md>), [algorithm](<https://devfeed.tech/tags/algorithm.md>), [algorithms](<https://devfeed.tech/tags/algorithms.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [awards](<https://devfeed.tech/tags/awards.md>), [awards-honors-and-fellowships](<https://devfeed.tech/tags/awards-honors-and-fellowships.md>), [binoy-kurien](<https://devfeed.tech/tags/binoy-kurien.md>), [computer-chips](<https://devfeed.tech/tags/computer-chips.md>), [cybersecurity](<https://devfeed.tech/tags/cybersecurity.md>), [dod](<https://devfeed.tech/tags/dod.md>), [invention](<https://devfeed.tech/tags/invention.md>), [juliana-furgala](<https://devfeed.tech/tags/juliana-furgala.md>), [laboratory](<https://devfeed.tech/tags/laboratory.md>), [lincoln-laboratory](<https://devfeed.tech/tags/lincoln-laboratory.md>), [magnetite](<https://devfeed.tech/tags/magnetite.md>), [melissa-choi](<https://devfeed.tech/tags/melissa-choi.md>), [military-inventions](<https://devfeed.tech/tags/military-inventions.md>), [military-technology-transfer](<https://devfeed.tech/tags/military-technology-transfer.md>), [mit-intellectual-property](<https://devfeed.tech/tags/mit-intellectual-property.md>), [mit-lincoln-lab](<https://devfeed.tech/tags/mit-lincoln-lab.md>), [operating-system](<https://devfeed.tech/tags/operating-system.md>), [platform](<https://devfeed.tech/tags/platform.md>), [quantum](<https://devfeed.tech/tags/quantum.md>), [quantum-computing](<https://devfeed.tech/tags/quantum-computing.md>), [quantum-enabled-computation-group](<https://devfeed.tech/tags/quantum-enabled-computation-group.md>), [quantum-microsystems](<https://devfeed.tech/tags/quantum-microsystems.md>), [r-d-100-awards](<https://devfeed.tech/tags/r-d-100-awards.md>), [rabindra-das](<https://devfeed.tech/tags/rabindra-das.md>), [satellites](<https://devfeed.tech/tags/satellites.md>), [secure-resilient-systems-and-technology-group](<https://devfeed.tech/tags/secure-resilient-systems-and-technology-group.md>), [system](<https://devfeed.tech/tags/system.md>), [tactical-edge-communications-group](<https://devfeed.tech/tags/tactical-edge-communications-group.md>), [technologies](<https://devfeed.tech/tags/technologies.md>), [u-s-armed-forces](<https://devfeed.tech/tags/u-s-armed-forces.md>)

### AI overview

Lincoln Laboratory's three 2026 R&D 100 Award technologies target embedded-system security, chip interconnection, and 5G interference. Magnetite builds on the formally verified seL4 microkernel, uses Rust services and least-privilege process isolation, and is being integrated into government space and flight platforms. The Quantum Microsystems platform links superconductor and semiconductor chips at wafer scale using AI-guided routing, custom lithographic wiring, and replaceable chiplets to reduce cabling and cryogenic wiring overhead; it is intended for transition to government and commercial foundries. A 5G canceler decodes standardized base-station transmissions, reconstructs interference including multipath distortion, and subtracts it per antenna channel. In tests with an over-the-air Nokia base-station signal, it restored a buried radar signal to within 2 dB of its original signal-to-noise ratio, preserving 89% of interference-free detection range. The Department of War is evaluating the algorithm.

### Source excerpt

A secure operating system, a quantum microsystem platform, and a 5G interference-canceling algorithm are named among the year's most significant innovations.

## Free Lunch Uncertainty: Sequence-Level R\\'enyi Entropy from Particle Decoding for Adaptive LLM Reasoning

DevFeed: [Free Lunch Uncertainty: Sequence-Level R\\'enyi Entropy from Particle Decoding for Adaptive LLM Reasoning](<https://devfeed.tech/articles/free-lunch-uncertainty-sequence-level-r-enyi-entropy-from-particle-decoding-for-adaptive-llm-reasoning-128310.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.13414>)

Author: Erfan Baghaei Potraghloo, Milad Heidari Sorkhehdizaj, Shahab Rahimirad, Aliasghar Pourghani, Seyedarmin Azizi, Souvik Kundu, Abolfazl Hashemi, Joseph Campbell, Vijay Gupta

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

Content type: article

Language: en

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

Topics: [llm-reasoning](<https://devfeed.tech/topics/llm-reasoning.md>), [Language models](<https://devfeed.tech/topics/language-models.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [Decoding](<https://devfeed.tech/topics/decoding.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [adaptive](<https://devfeed.tech/tags/adaptive.md>), [adaptive-compute](<https://devfeed.tech/tags/adaptive-compute.md>), [all](<https://devfeed.tech/tags/all.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [benefit](<https://devfeed.tech/tags/benefit.md>), [best](<https://devfeed.tech/tags/best.md>), [budget](<https://devfeed.tech/tags/budget.md>), [completion](<https://devfeed.tech/tags/completion.md>), [compute](<https://devfeed.tech/tags/compute.md>), [consistency](<https://devfeed.tech/tags/consistency.md>), [cost](<https://devfeed.tech/tags/cost.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>), [decoding](<https://devfeed.tech/tags/decoding.md>), [entropy](<https://devfeed.tech/tags/entropy.md>), [free](<https://devfeed.tech/tags/free.md>), [llm](<https://devfeed.tech/tags/llm.md>), [llm-reasoning](<https://devfeed.tech/tags/llm-reasoning.md>), [reasoning](<https://devfeed.tech/tags/reasoning.md>), [sequence](<https://devfeed.tech/tags/sequence.md>), [uncertainty](<https://devfeed.tech/tags/uncertainty.md>)

### AI overview

Sequential Monte Carlo (SMC) decoding for power sampling can estimate sequence-level Rényi entropy (SRE) at zero marginal cost, with a guarantee at every particle budget. A Rényi chain rule also yields stepwise entropies: orders above one are tied to each step's most likely token and track log-probability, while the order-1/2 measure accumulates the log of per-step search effort. That order was the strongest zero-cost selective-prediction signal in 10 of 13 settings. On MATH500, its AUROC exceeded log-probability by 0.022-0.044 across four models, including within length quartiles. An 8-particle probe identified problems benefiting from more particles; a two-pass policy matched 64-particle accuracy using 34.2% fewer particles over three seeds.

### Source excerpt

arXiv:2611.13414v1 Announce Type: cross Abstract: Quantifying uncertainty over an entire LLM completion is critical for selective prediction and adaptive compute. Zero-cost signals such as mean token entropy and log-probability are local to tokens or to one completion, and self-consistency needs many extra completions. We show that Sequential Monte Carlo (SMC) decoding for power sampling already estimates the normalizing constant of the power distribution, which gives the sequence-level R'enyi entropy (SRE) of the model's completion distribution at zero marginal cost, with a guarantee at every particle budget. A R'enyi chain rule extends this readout to a family of stepwise entropies of any order. We prove that orders above one are tied to the most likely token of each step, which explains why they track log-probability, while the order-1/2 member accumulates the log of per-step search effort. This member is the best zero-cost selective-prediction signal in 10 of 13 settings. On MATH500 its AUROC is 0.022 to 0.044 above log-probability with all four models and within length quartiles, and the same order is selected when each setting is held out. An 8-particle probe identifies the problems that benefit from more particles, and a two-pass policy matches 64-particle accuracy on MATH500 with 34.2% fewer particles over three seeds.

[Next page](<https://devfeed.tech/topics/decoding.md?cursor=WyIyMDI2LTExLTExVDA1OjAwOjAwKzAwOjAwIiwgImUyMDIyYWZjLTUyNjItNDRlZi05MzhhLWEzZjQ0YmIwOGM1OCJd>)