Specialty
A preprint evaluates G-MARK, a provenance-aware knowledge graph for cooperative-driving reasoning. Reported gains were strongest in occlusion, hidden-object and motion tasks, while communication use in the future-trajectory comparison was 0.0159 MB per sample.
Scientific publication3 min read
An arXiv preprint reviews selected neural-computation literature and identifies a forward–backward disconnect: the audited configurations span several kinds of forward dynamics, but scalable learning evidence clusters around global or gradient-derived error propagation. The review warns that its counts are descriptive, not estimates of field-wide prevalence.
Scientific publication4 min read
An arXiv preprint reports that SATS led several benchmark comparisons, including tests on datasets held out from pretraining, while using fewer parameters than key baselines.
Scientific publication3 min read
A retrospective comparison found different strengths between two forecasting models: Chronos-2 had the lowest aggregate error, while TabPFN-TS was better calibrated and remained among the leading models in a second network.
Scientific publication4 min read
A preprint testing FedCurv-DR, a method for federated continual learning, found higher final accuracy and less-negative forgetting scores than FedAvg in a simulated WikiArt benchmark. FedCurv-based methods also showed lower client-level disparity, while FedAvg used the least measured energy.
Scientific publication4 min read
A methods preprint found that the strongest early time-series classifier changed with the type of controlled data drift.
Scientific publication4 min read
An arXiv preprint describes a dual-stream forecasting model that reports accuracy and efficiency gains on benchmark data, while leaving broader questions about generalisation, calibration and interpretability unanswered.
Scientific publication7 min read
A modeling study reports that smaller proxy runs can inform learning-rate choices for much larger mixture-of-experts models. Retrospective checks were close, but the predicted setting for a 10-trillion-token run was not tested in a full-scale sweep.
Scientific publication5 min read
A methods preprint compares a JEPA with separate prediction branches against standard and other benchmark baselines across five systems, with results favoring the factorized design within the tested settings.
Scientific publication3 min read
A new machine-learning method adds geographic information to sparse autoencoders and turns their internal features into rule-based explanations. The preprint reports close computational agreement between those rules and the underlying features, while leaving expert validation and operational use untested.
Scientific publication3 min read
A computational preprint describes inference-time methods that steer a discrete diffusion language model toward sequence-level rewards without retraining. Its main comparison favored the nested methods on the paper’s automated measures, within a narrow evaluation.
Scientific publication4 min read
A timing-focused neural-circuit model reproduced reported circuit behaviors, but its measured jitter varied substantially between platforms.
Scientific publication3 min read
A relation-first token-mixing design recorded lower final-validation NLL than matched multi-head attention at approximately 10 million, 30 million and 100 million parameters. Its fused implementation was faster than a materialized reference but remained slower than FlashAttention in scale-matched production workloads.
Scientific publication3 min read
An arXiv preprint presents a deterministic Newton-polygon algorithm for exact local RLCTs in rational bivariate polynomials and contact-equivalent models. In a symbolic benchmark, the exact calculation was faster than SGLD at two reported timings, but the method’s scope remains limited beyond two parameters.
Scientific publication5 min read
An arXiv preprint reports measurable word-level information in EEG during silent reading. The result included both context-tracking and context-independent components, but it came from one participant and used validation-selected scores.
Scientific publication3 min read
A methods preprint proposes pooling related groups to estimate a shared pattern, then modeling each group’s offset. Its reported comparisons show lower mean squared error than pooled neural networks in selected simulations, Beijing PM2.5 prediction and UTKFace age prediction.
Scientific publication3 min read
An arXiv preprint describes DICS, a method that narrows the candidate splits considered during classification-tree training. The reported speedups were substantial in synthetic tests and remained visible on benchmark data, but the findings are limited to tested implementations and classification tasks.
Scientific publication3 min read
A modeling analysis of 105 clinical HSAT recordings found that some learned sleep-signal relationships persisted across groups, while others differed by sex or age. The findings generate hypotheses but do not establish cause and effect.
Scientific publication3 min read
A new arXiv methods preprint proposes confidence sets and active endpoint bracketing for sparse models whose dictionary atoms are highly coherent. Synthetic tests found that the method generally matched exhaustive finite-bank conclusions while avoiding finer reports the evidence did not support.
Scientific publication4 min read
A controlled preprint comparison of three ceiling-mounted sensing systems found a trade-off between detailed activity recognition and robustness to changing room layouts. IR-UWB led in cross-participant testing, FMCW led in the strictest layout holdout, and all three performed strongly on coarse sleep-disruption monitoring.
Scientific publication6 min read
A preprint evaluates BERT-LER, a transformer model for structured electronic health records that represents laboratory tests and their values. It reports higher benchmark scores in the evaluated tasks, while the study’s retrospective data and qualitative explanation checks limit what can be concluded about clinical use or cause.
Scientific publication3 min read