Machine learning news — deep learning, reinforcement learning, neural architecture search, diffusion models, and new ML frameworks and libraries.
Researchers propose a structural pruning framework for Mixture-of-Experts (MoE) models that reformulates prune-ratio allocation as a channel-score coverage maximization problem, solved via attribution-based approximation…
Researchers introduced ForecastBench-Sim, a simulated-world forecasting benchmark built on game rollouts from Freeciv, to evaluate AI systems' probabilistic reasoning under dynamic world states. The benchmark generates c…
Researchers propose Data-Forcing Distillation (DFD), a post-training framework that restores diversity and fidelity in few-step video generation models by addressing mode collapse and over-saturation artifacts. DFD achie…
Researchers introduced ARIADNE, a training-free routing framework for dynamic adapter selection in parameter-efficient fine-tuning ecosystems. It selects task-specialized adapters at inference time by measuring input pro…
Researchers propose RODS (Reward-driven Online Data Synthesis) to address the depletion of informative samples in multi-turn tool-use reinforcement learning. RODS uses progress reward variance as a boundary detector to c…
Researchers propose ThinkDeception, a novel framework that uses multimodal large language models and chain-of-thought reasoning to detect deception with interpretability. It introduces a progressive training strategy and…
Researchers introduced VISUALSKILL, a multimodal skill library for computer-use agents that combines text and visual figures to improve performance on long-horizon tasks. In tests, the system achieved a 0.456 average sco…
Researchers introduced ThousandWorlds, an ML-ready benchmark dataset for emulating exoplanet climates using machine learning, containing approximately 1800 simulations from five global climate models. The benchmark aims …
A new study comparing Vision Transformers and CNNs for face presentation attack detection found that pretrained Vision Transformers reduce demographic bias and improve accuracy. The DeiT-S model achieved 97.27% accuracy …
Researchers propose using Stable Diffusion to generate synthetic indoor scenes for data augmentation, improving indoor scene recognition models. Experiments on the MIT Indoor Scene dataset show enhanced training with lim…
Researchers propose RGNet, a neural network architecture inspired by the renormalization group, to address class imbalance and noise in fault diagnosis. The model hierarchically compresses feature spaces and introduces R…
Researchers introduced Ghost Attractor Networks, a dynamical decoder that uses basin-structured latent representations for closed-loop sequential generation. The 2.3-million-parameter Ghost matched the offline accuracy o…
Researchers introduced a learned generalization of phase correlation that lifts the restriction of fixed basis, enabling direct measurement of inter-image relationships in the Fourier domain for dense non-rigid deformati…
A new study on arXiv presents an experimental analysis of neural network-based image classification on the CIFAR-10 dataset, comparing fully connected and convolutional architectures. The convolutional network achieved 7…
A team led by James J. Collins at the Wyss Institute used a deep learning pipeline to identify two novel compounds active against Neisseria gonorrhoeae, the bacterium causing gonorrhea. The compounds, discovered by train…
QuantaMind built the 'Quant Audit' feature to measure how quantization degrades AI agent reasoning, warning that leaderboard scores misrepresent real-world performance. The tool systematically evaluates performance drop-…
A developer in Timor-Leste built FurakToon, an AI cartoon generator that turns sentences into anime or cartoon images in seconds, in about two weeks for the Mind the Product 'Everyone Ships Now' hackathon. The app suppor…
Researchers introduced Regressive Plasticity Schedule (RPS), a two-stage post-training schedule that couples a learning-rate drop with a curriculum boundary between easier and harder data. Testing on Qwen3-8B with ARC-AG…
A developer building WearEdge Pro, a wearable industrial edge AI runtime, tested five small multimodal models on a Jetson device to find the best baseline for an industrial edge agent. Gemma 4 E2B emerged as the best pro…
Intel has published the x86 AI Compute Extensions (ACE) specification, defining new matrix multiplication primitives and reduced precision data formats to accelerate machine learning workloads. The extensions integrate t…