cs.NE

Neural and Evolutionary Computing

Covers neural networks, connectionism, genetic algorithms, artificial life, adaptive behavior. Roughly includes some material in ACM Subject Class C.1.3, I.2.6, I.5.

pith recommended most recent

cs.NE arXiv:2610.16102

Tacit knowing is an operating regime of a memory

The same hyperdimensional memory yields exact audit or emergent path valuation by flipping one parameter, explaining the tacit-explicit…

By Hiroyuki Chuma, Kanji Otsuka, Yoichi Sato and 1 more · “Where Polanyi Meets Galois: Tacit Knowing as an Algebro-Deterministic Operation, Toward Re-rooting AI-Generated Science”

arXiv submitted 2026-10-12 · Pith reviewed 2026-10-14 Pith rank: recent upvotes, halved daily

open re-runnable review →

cs.LG arXiv:2610.18085

No optimizer keeps rescaled Transformers on the same trajectory

Theoretical proof and practical rebalancing show why weight scale matters and how to fix it

By Bum Jun Kim · “Same Function, Different Trajectory: No Standard Optimizer Is Equivariant to Transformer Reparameterization”

arXiv submitted 2026-10-13 · Pith reviewed 2026-10-14 Pith rank: recent upvotes, halved daily

open re-runnable review →

cs.NE arXiv:2610.22211

Census-timed restarts beat fixed schedules for genetic algorithms

Mark-recapture counting tells a GA the search region is empty; on three NP-hard benchmarks it beats fixed restarts.

By Anton Eremeev, Yulia Zakharova · “A restart rule for genetic algorithms based on Schnabel census”

arXiv submitted 2026-10-15 · Pith reviewed 2026-10-16 Pith rank: recent upvotes, halved daily

open re-runnable review →

One attention step retrieves archetypes the layer never saw

cond-mat.dis-nn arXiv:2610.22191 reviewed 2026-10-16

Gain sets a Hamming radius on spiking hardware: 238/256 exact

cs.NE arXiv:2610.22083 reviewed 2026-10-16

Weight perturbation hits 97% on MNIST

cs.LG arXiv:2610.21665 reviewed 2026-10-16

Multitask slashes multi‑objective optimization runtime to match the easier problem

cs.NE arXiv:2610.21311 reviewed 2026-10-16

Analytical latency bounds for spiking neural networks

cs.NE arXiv:2610.19732 reviewed 2026-10-15

Higher temporal representation boosts network performance on time-based tasks

cs.NE arXiv:2610.19730 reviewed 2026-10-15

Projection turns three-body neural interactions into pairwise

q-bio.NC arXiv:2610.19487 reviewed 2026-10-15

CTM follows a shared fixed-point curve for all ImageNet

cs.LG arXiv:2610.19220 reviewed 2026-10-15

Saliency-picked patches beat full-image replay on tight budgets

cs.LG arXiv:2610.17998 reviewed 2026-10-14

Dimensional consistency stabilizes LLMs across 16× learning rates

cs.LG arXiv:2610.17969 reviewed 2026-10-14

Interaction stabilizes collective decisions and scales efficiently

physics.soc-ph arXiv:2610.17506 reviewed 2026-10-14

Newton's method parallelizes spiking LIF neuron training

cs.NE arXiv:2610.16924 reviewed 2026-10-14

RL with temporal structure boosts MI-BCI decoding by 2–3 points

cs.NE arXiv:2610.16921 reviewed 2026-10-14

Direction-aware graph transformer beats NAS predictors by wide margins

cs.LG arXiv:2610.16822 reviewed 2026-10-14

Optimised fuzzy spatial clustering now in R with seven metaheuristics

cs.NE arXiv:2610.16356 reviewed 2026-10-14

LLM-guided structural control boosts conflict graph search

cs.NE arXiv:2610.16158 reviewed 2026-10-14

Adam statistics reveal hidden retraining susceptibility in representations

cs.LG arXiv:2610.15830 reviewed 2026-10-14

browse all of cs.NE →

How this page is ordered

Papers reviewed in the last 7 days lead, then the papers readers read most. The order is not a quality score.

Computer Science · part of cs · all fields

Most relied on in cs.NE

arXiv papers in this field that the most other arXiv papers use, build on, compare with or cite as support, read from their citing sentences.

Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling 864 rely on it · 771 use its method · 6 build on it · 3 contest it
Progressive Growing of GANs for Improved Quality, Stability, and Variation 581 rely on it · 181 use its method · 5 build on it · 1 contest it
Improving neural networks by preventing co-adaptation of feature detectors 506 rely on it · 472 use its method · 3 build on it
Searching for Activation Functions 447 rely on it · 413 use its method · 3 build on it · 2 contest it
Network In Network 363 rely on it · 322 use its method · 3 build on it
Deep Learning using Rectified Linear Units (ReLU) 319 rely on it · 316 use its method
A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networks 239 rely on it · 126 use its method · 2 contest it
Generating Sequences With Recurrent Neural Networks 237 rely on it · 205 use its method · 5 build on it · 2 contest it