Neural Networks

Researchers at the University of Cambridge and Geodesic Research have proposed a new framework called Developmental Cognitive Interpretability (DCI) that models how an AI system's motivations, goals, and intentions chang…

EV-QA-Framework, an open-source machine learning tool for electric vehicle battery quality assurance, has been released on GitHub. The framework combines rule-based validation with Isolation Forest anomaly detection and …

A developer has proposed a dual-mode cognitive architecture for AI systems, combining a fast "Intuition" mode based on attention vectors with a deeper "A11" reasoning engine that activates for complex or high-stakes task…

A fractal time generator from the Krestianstvo Wavefront Evaluator has been substituted into the Schrödinger equation as a quantum kinetics operator, bridging quantum wave mechanics with Hopfield associative memory. The …

A new 306M-parameter language model architecture, the Cognitive Categorical Transformer (CCT), achieved 21.27 validation perplexity on WikiText-103, a 12% relative improvement over a fine-tuned GPT-2 Small baseline. The …

Researchers found that a world model trained on random physical exploration develops spatial semantic structure in its latent space that mirrors physical geometry, without any linguistic supervision. The model achieved a…

Researchers at Qwen2.5-3B-Instruct found that supervised fine-tuning (SFT) causes greater disruption to internal computational circuits and more catastrophic forgetting than reinforcement learning (RL) when adapting larg…

Researchers at arXiv have discovered that knowledge editing methods ROME and MEMIT, which modify factual associations in transformer models, rely on a shared functional mechanism rather than fact-specific weight changes.…

Researchers have developed Deep Visual Coding, a frequency-domain representation inspired by 1990s image codes that creates psychovisual-style abstractions for deep learning. The approach enables the first psychovisual-b…

Researchers have developed GeRaF 2.0, a neural reconstruction framework that uses radio frequency signals to create 3D models of objects hidden behind walls or other occlusions. The system integrates visible line-of-sigh…

Researchers have developed GeRaF, the first neural implicit learning method for reconstructing 3D geometry from radio frequency signals, enabling see-through-occlusion sensing without relying on RGB or LiDAR. The system …

Researchers have developed a self-supervised deep learning framework that enhances Sentinel-1 Stripmap imagery by exploiting azimuth subaperture decomposition, generating paired training data without external sensors or …

Researchers have identified a critical underfitting problem in diffusion bridge models near the target endpoint, caused by a mismatch in noise levels between network input and regression target. The team proposes the Noi…

A new study decomposes the Platonic Representation Hypothesis into signal, bias, and noise, finding that cross-modal alignment in AI models stems from a universal linear encoding of object-attribute relationships. Resear…

Researchers have reframed context distillation as a latent memory management problem, distilling each context into an independent LoRA adapter to form a modular memory bank. The framework retrieves candidate memories for…

Researchers at an undisclosed institution analyzed LoRA fine-tuning in Gemma-2-9B using sparse autoencoders, finding that adapter-specific feature dictionaries show weak geometric alignment with pretrained features acros…

Researchers introduced Spectral Guidance, a framework that controls diffusion models by exploiting the geometric structure of noise corruption to identify a small set of informative features. The method projects guidance…

Researchers developed a mathematical framework for sequential surrogate modeling of three-phase black-oil reservoir dynamics using Fourier Neural Operators (FNO) and physics-informed neural operators (PINO), applied to t…

Researchers have introduced CosmicFish-HRM, a compact language model that uses a Hierarchical Reasoning Module (HRM) to dynamically allocate computational effort based on input complexity rather than applying fixed compu…

Researchers introduced behavior-aware auxiliary corrections for off-policy temporal-difference learning, replacing the standard covariance matrix with the behavior Bellman matrix in the TDC and TDRC algorithms. The new B…