Neural Networks

A brain imaging study found that experienced psychedelic users process threatening facial expressions more efficiently than nonusers, with faster and more accurate recognition of angry faces. The research, led by Paweł O…

A brain-computer interface (BCI) developed by UC Davis, Brown University, and Mass General Brigham enabled a man with ALS to communicate independently at home for nearly two years, achieving 99% word accuracy and a speed…

A blog post derives the backward pass of attention mechanisms step by step, using a concrete example to illustrate gradient computation for Q, K, and V matrices. The derivation builds on FlashAttention and FlashAttention…

Researchers from UNSW, University of Sydney, and UTS will present at the Australian AI Safety Forum 2026 on how neurotechnology could impact AI safety, exploring the 'neural democratisation of AI' hypothesis. The talk wi…

Researchers propose replacing the inner product score in sparse autoencoders with a learned blend of cosine similarity and input magnitude, finding that cosine-scored SAEs learn more human-recognizable features and avoid…

Researchers used machine learning models to predict exam outcomes from physiological data such as electrodermal activity, heart rate, and skin temperature. Deep learning models like transformers and LSTMs captured comple…

A new study compares 19 graph neural network layer types for driving trajectory prediction, finding that ARMA, Chebyshev, and topology-aware layers outperform others. The research provides design principles such as sum-b…

Researchers introduced UtVAA, an ultra-tiny Vision Transformer architecture with a novel Affix Attention block for mobile image classification. The smallest variant has 204.67K parameters and 53.95M FLOPs, achieving comp…

Researchers introduced the Separable Neural Architecture (SNA), a function representational class combining neural approximation with tensor decomposition, which decouples localized coordinate functions from global inter…

Researchers introduced GRAPE, a training framework that progressively exposes neural network parameters to improve adversarial robustness. On CIFAR-10 with ResNet-18, GRAPE achieved 56.94% PGD-20 robust accuracy, a 5.24 …

Researchers introduced a hierarchical model for ball action anticipation in football broadcast video, achieving 17.91% mAP on the SoccerNet benchmark. The system uses a GRU and Transformer decoder with input-conditioned …

Researchers achieved 98.91% accuracy in multimodal emotion recognition from physiological signals by combining LSTM, TCN, and Transformer models with late-fusion ensemble on the WESAD dataset. Transformer models excelled…

Researchers from an undisclosed institution introduced Convolutional Nearest Neighbors (ConvNN), a unified framework that shows convolution and self-attention are special cases of k-nearest neighbor aggregation. ConvNN c…

Researchers introduced a geometric framework to identify 'AI engrams'—identifiable memory traces in deep neural networks analogous to biological memory units. The method enables surgical manipulation of learned knowledge…

Neural resonance theory (NRT) proposes that musical preferences are innate and biological, driven by the brain's physical resonance with sound frequencies. According to NRT, the nervous system syncs to music, favoring si…

A developer explains that a neural network is not magic but a mechanical process of weighted detectors stacked in layers, turning raw input into higher-level features to make a decision. The network starts with random we…

Researchers at Hiroshima University and Nagoya University developed a machine learning tool called SPERRFY that decoded how gene activity patterns guide neural wiring across the mouse brain. The study, published in PNAS,…

A single dose of psilocybin, the active ingredient in magic mushrooms, induced anatomical brain changes detectable a month later in healthy first-time users, according to a study by researchers at Imperial College London…

A developer has built a scientific computing platform called Cognitive Discovery System entirely in pure Python without NumPy, achieving O(N log N) FFT, O(N³) LU decomposition, and a quantum state-vector simulator with O…

A developer explains how the Transformer architecture works, from self-attention to modern LLMs. The key innovation is that Transformers compare tokens directly via attention rather than processing sequences step by step…