Machine Learning News

TinyCardioUNet, a lightweight UNet from researchers Seungwoo Han, Ingon Chanpornpakdi, Motoi Noda, Puwadej Leelasiri, Ibuki Hiruma and Toshihisa Tanaka, reconstructs electrocardiography (ECG) from a chest-worn six-axis i…

Researchers proposed SpiKFAX, a second-order optimization method that builds a computationally tractable, Kronecker-factored approximation of the Fisher information matrix specifically adapted to spiking neural networks …

Researchers introduced Direct Message Approximation (DMA), a consistency-based framework that approximates factor-to-variable messages directly on factor graphs rather than approximating marginals as expectation propagat…

A new arXiv paper introduces the Spectral Concentration Ratio (SCR), an offline sensitivity proxy that assigns each residual branch a deterministic lifetime so pretrained diffusion models can skip recomputing frozen unit…

A 27B model trained to predict proof difficulty outperforms frontier general-purpose models on that task and, when optimized for a theorem-interestingness metric defined as the ratio of proof length to statement length, …

Researchers Samuel McCandlish and colleagues submitted "Scaling Laws for Neural Language Models" to arXiv on 23 January 2020, showing that language model cross-entropy loss falls as a power-law with model size, dataset s…

A new benchmark called ExplorationBench measures how well AI systems explore in verifiable alien worlds, according to the benchmark's authors, who frame scientific discovery as beginning where known problems end. The aut…

The Massachusetts Gaming Commission ordered its staff to engage directly with DraftKings to determine how the company's artificial intelligence targets bettors, following a September 19 New York Times investigation into …

TypeSafe AI released Jev, which it calls the first "System One model," a non-autoregressive model that answers typed structured questions in a single forward pass with a probability attached to every answer, priced at $0…

A developer released vqc-rl, a Tk-based GUI that visualizes variational quantum circuits (VQC) used as the action-value function in deep Q-learning on a stochastic 4×3 grid world, built from a 2024 internship at Porto Un…

Microsoft published a Nature paper this week describing RetroChimera, a predictive model that accelerates chemical synthesis so researchers can explore a wider range of custom-made molecules for medicine, materials, and …

Two September 2026 papers propose post-training alignment techniques that expose models to misaligned contexts and then enforce aligned targets: ARCF, which reduced unsafe generations on a benchmark of 5,000 harmful prom…

A University of Cambridge and Google DeepMind team published XConf on 15 September 2026, a confidence-estimation method that reads confidence from a bank of past graded episodes rather than from the model's current answe…

A developer argues that most AI agent evaluation pipelines rely on proxy metrics like accuracy, precision, recall and reward-model scores that measure what an agent did rather than what actually happened, allowing agents…

DuckDB can query Hugging Face Hub datasets directly over its hf:// protocol without downloading them, an integration added in DuckDB v0.10.3, released May 22, 2024, and announced in May 2024. The scheme, built on DuckDB'…

Builder.io's first Figma-to-code system used a gradient-boosted decision tree model (XGBoost) that compared pairs of candidate node merges, feeding each candidate 21 geometry and structure features for 42 inputs total re…

AMD detailed model weight profiles that break down how many parameters go to embeddings, attention, and dense layers, explaining why the amd/Llama-3.1-8B-Instruct-FP8-KV checkpoint is 9.08 GB rather than the roughly 8 GB…

MimicScribe's live-preview transcription, built on a batch ASR model with overlapping-window merging, transcribed 18.8 hours of meetings and earnings calls with 15.5% of words wrong and 96.9% of spoken numbers correct, w…

Researchers behind the paper "Evolutionary Architecture Search through Grammar-Based Sequence Alignment" (arXiv:2512.04992) use sequence alignment to identify the edits connecting two neural architectures and then draw o…

Iluvatar Labs released Kev, an open-source tool on GitHub that converts any LLM into a decision model returning choices and probabilities instead of generated text, starting with Qwen3's 1.7B, 4B, and 8B models. On a 276…