The IG-GAN model, which leverages intrinsic geometry and Bézier surfaces to generate aerodynamic data, achieves a 97.41% reduction in predicted Mean Squared Error (MSE) for velocity prediction on the …
The Associative Recurrent Memory Transformer (ARMT) extends language model context length efficiently, using 30% fewer FLOPs and constant memory scaling while generalizing to out-of-distribution conte…
AI-driven summaries risk stripping away critical insights needed for informed decision-making, according to a new analysis. The reliance on executive summaries, fueled by AI's ability to condense info…
Researchers have introduced Weight-Adjusted Gradients (WAG), a novel method for estimating parameter importance in Large Language Models that reveals a tiny subset of parameters disproportionately inf…
RecursiveMAS, a new framework for multi-agent AI collaboration, boosts accuracy by 8.3%, improves inference speed by up to 2.4 times, and reduces token usage by 34.6% to 75.6% across nine benchmarks i…
A new study reveals that existing class unlearning methods in machine learning ignore class geometry, enabling Class Membership Inference Attacks (CMIA) that leak information about supposedly forgotte…
A new study on robust AI planning introduces non-asymptotic convergence bounds for Q-learning and actor-critic algorithms in distributionally strong Markov Decision Processes, achieving an optimal str…
Researchers have discovered that deep neural models for click-through rate predictions suffer a sharp decline in validation performance after just one training epoch, even as training loss improves, d…
A new neural network optimization method using a symmetric-exponential weight reparameterization achieved up to 30% faster training on transformers, reaching equivalent validation loss in 1.32 to 1.49…
A new feedback-driven approach enhances Retrieval Augmented Generation (RAG) systems by integrating human input to iteratively improve response quality, according to a recent study. The method adds an…
Researchers have introduced STEC, a framework that improves multi-hop question answering by compressing search data into evidence representations for each candidate answer and then verifying which ans…
A new regulation-driven fine-grained hierarchical classification framework has been proposed to improve accuracy and transparency in compliance tasks such as customs tariff classification and export c…
CLIR-Bench, a new benchmark built from de-identified ICU records with 6,600 QA instances across 11 clinical variables, reveals that existing generalist AI models struggle to reason over irregular clin…
A major Eastern European bank has deployed a transformer-based foundation model that unifies multimodal user event sequences into a single chronological stream, outperforming traditional predictive mo…
Researchers found that the general-purpose language model LLaVA 1.6 can simulate speech errors of aphasia patients with high accuracy, matching six of seven error types for 97.8% of 278 individuals an…
Researchers have found that the multimodal language model LLaVA 1.6 can replicate six of seven categories of naming errors seen in stroke-induced aphasia, matching real-world clinical data in 97.8% of…
Chinese AI laboratory DeepSeek, which closed a $7 billion funding round weeks ago, is already seeking additional capital to maintain its aggressive pricing strategy and expand infrastructure for custo…
Google DeepMind CEO Demis Hassabis called for the US to establish a robust frontier AI model review process, warning that artificial general intelligence (AGI) is "probably only a few short years away…
A new research framework shows that integrating personal visual memory into AI systems improves tool selection from images, with tool-query relevance dropping by 0.47 points (11.2%) and overall utilit…
A study on on-device English-to-Traditional-Chinese subtitle translation shows that reducing the vocabulary of the LMT-60-0.6B model from 151,000 to 64,000 tokens improves speed and privacy. The Local…