Neural Networks

Researchers at the Weizmann Institute of Science developed Brain-IT, an AI system that reconstructs images a person is viewing from their brain scans and needs only 60 minutes of scan data to adapt to a new person, versu…

Developer krypticlogan released ZGC, a tensor graph compiler written in Zig, on GitHub after pivoting from a specialized neural network graph runtime. ZGC targets Zig versions 0.16.0 with 0.17.0 support planned, and the …

Researchers Samip Dahal, Bishwas Mandal, Serdar Gülbahar and Akshay Vegesna released Dust, a zeroth-order pretraining method that perturbs activations independently at every token so one forward pass evaluates a virtual …

A Stochastic Blog post builds a multilayer perceptron on raw MNIST pixels, verifying a hand-derived gradient against autograd before running a controlled sweep over seven activations, four initializations, three normaliz…

A new arXiv paper (2610.02253v1) establishes approximation rates for dropout ReLU networks, showing that the unit ball of W^{n,∞}([0,1]^d) can be approximated uniformly by constant-depth networks of size Õ_{n,d}(p^{-9}ε^…

A study posted to arXiv (2610.02284v1) found limited support for the hypothesis that natural interpulse-interval (IPI) variation contributes substantially to automated bat-species classification, with transformer model P…

A new arXiv paper (2610.02344v1) develops an early-training stability theory for Joint-Embedding Predictive Architectures (JEPAs) that explains representation collapse through a per-mode stability ratio μ_i = γ_i / σ_i, …

Princeton researchers introduced SCION, a hierarchical compositional scene representation that replaces millions of independent 3D Gaussians with a compact vocabulary of reusable primitives and lightweight world-space in…

A new arXiv paper, arXiv:2610.02494v1, proposes DeepStratNet, a context-aware coordinate regression framework that reformulates 3D seismic horizon tracking as a bounded coordinate regression problem instead of dense sema…

IBM's NorthPole brain-inspired chip is more than 20 times as fast and roughly 25 times as energy efficient as any microchip currently on the market for AI tasks, according to an IBM study led by Dharmendra Modha, IBM's c…

Ukrainian software developer Rodion is building Messagram, a privacy-focused messenger that uses on-device face-tracking neural networks and the front-facing camera to detect when someone is peeping at a user's screen. A…

A Generative Adversarial Network (GAN) consists of two neural networks — a Generator that turns random noise into fake data and a Discriminator that outputs the probability a sample is real — trained simultaneously throu…

A from-scratch PyTorch implementation of a physics-informed neural network (PINN) recovers blood flow, viscosity, and wall shear stress in a narrowed artery from 40 noisy velocity readings, according to Towards Data Scie…

Researchers at Brown University led by engineering professor Arto Nurmikko built a brain-inspired imaging system that combines a dynamic vision sensor with a deep spiking neural network to image and track randomly moving…

A master's thesis benchmarked three deep learning architectures, one of them novel, across varying preprocessing pipelines for EEG motor imagery task classification on a consumer-grade OpenBCI EEG cap. The author is aski…

OpenTrainDNN, an open-source client-side web application from developer MatiwosKebede, runs real backpropagation, Adam optimizer weight updates, and gradient clipping entirely in the browser to visualize neural network t…

Researchers introduced SW-KAN, a Kolmogorov-Arnold Network architecture that uses Stieltjes-Wigert q-orthogonal polynomials defined on the semi-infinite domain (0, infinity), according to an arXiv paper (arXiv:2610.00050…

A small DeltaNet variant with fixed token-specific key biases and 32 KiB of recurrent matrix state reached 99.95% mean recall accuracy across three training seeds when remembering 32 new key-value pairings per sequence, …

Researchers proposed FedCORE, a federated adaptation framework that represents multimodal Encoder and graph neural network (GNN) updates through a shared low-dimensional latent state, according to an arXiv paper (arXiv:2…

Researchers proposed STCFormer, an adaptive spatio-temporal Transformer that dynamically groups weather stations by their local evolution within each temporal patch, and reported the lowest 24-hour mean squared error on …