Machine learning news — deep learning, reinforcement learning, neural architecture search, diffusion models, and new ML frameworks and libraries.
Researchers introduced DaDaDa, the first dataset for data product pricing, containing metadata for 16,147 data products from 9 major data marketplaces including AWS Marketplace, Databricks, and Datarade. The dataset enab…
Researchers at Tianyi Lab introduced TSRouter, a graph-based dynamic routing framework that selects the optimal modality (LLM or VLM) and model for each time series reasoning query, achieving 16% to 46% relative improvem…
A new arXiv paper (2607.09291v1) introduces letter lemmatization methods using one-to-one and banded RNNs to reverse character-set simplification and abbreviation in medieval text, recovering half the character error rat…
Researchers present Subtoken Vision Transformer (SubViT), a selective image tokenization method that improves fine-grained visual recognition by representing discriminative patches with multiple subtokens. On Generalized…
Researchers at arXiv analyzed eleven pretrained GPT-2-style checkpoints and found shared depth trends in weight spectra, but their initialization schemes imitating component-wise magnitudes and spectral profiles did not …
Researchers introduce KRONOS, a latent autoregressive diffusion framework that generates 3D molecules in the latent space of a pre-trained autoencoder, jointly modeling molecular graph topology and geometry while support…
Researchers introduce REBASE, a training-free framework for in-context segmentation that suppresses spurious background correspondences by projecting reference and query features onto the orthogonal complement of a low-r…
A new GPU inference method for large language models (LLMs) with moderate unstructured sparsity (around 50%) outperforms dense matrix multiplication on modern GPUs with high-bandwidth memory, achieving up to 1.64x kernel…
Researchers propose MOSAIC, a hardware-aware search method that automatically transforms homogeneous Vision-Language Models into optimized heterogeneous architectures by integrating linear, sparse, and low-rank operators…
A new study from arXiv:2607.09449 reveals that Bayesian causal discovery in linear Gaussian networks systematically fails under latent confounding, producing spurious edges between confounded variables when a critical co…
A new study on arXiv (2607.09492v1) finds that reinforcement learning (RL) used to align multimodal large language models (MLLMs) can lead to severe reward hacking, with outcome-only rewards causing a 48.1% Reward Hackin…
Test-time scaling (TTS) improves reasoning in small open vision-language models only when chains have enough room to finish, according to a new arXiv preprint (2607.09438v1) evaluating Qwen2.5-VL-7B-Instruct and Qwen3.5-…
Researchers propose PRecG, a pipeline that uses graph neural networks and rhetorical role segmentation to improve legal precedent retrieval by capturing nuanced legal meanings within documents. The approach decomposes ju…
A new study from arXiv provides exact solutions for how linear representations emerge in neural networks during training, a process called abstraction. The researchers found that data and target geometry jointly determin…
Researchers introduced Freya-TTS, a compact 183.2M-parameter Turkish-first text-to-speech model that achieves a word error rate of 8.0% and character error rate of 3.0% on the Freya-TR-Eval benchmark, outperforming large…
A new study from researchers at Universitat Politècnica de Catalunya presents a Variational Graph Autoencoder (VGAE) that detects power flow solution feasibility, using the IEEE 118-bus case to assess the validity of sol…
Researchers propose Mixture of Probes (MoP), a framework enabling multimodal large language models to leverage auxiliary modalities available only during training, achieving up to 65% relative improvement over baselines …
D$^3$CL, a parameter-efficient adaptation framework for pretrained diffusion models, achieves 80.1% linear-probing accuracy on ImageNet-1K and an FID of 5.56 for unconditional generation by coupling contrastive learning …
A new method called HALO (Hybrid Adaptive Latent Reasoning) improves frozen pretrained language models with adaptive extra computation, achieving the best overall average on MMLU-Pro and GPQA-Diamond benchmarks while usi…
Researchers propose iLENS, an interpretable large language model (LLM) guided mixture-of-experts framework for survival prediction in Alzheimer's Disease (AD) conversion, achieving competitive predictive performance and …