Researchers have introduced BalDRO, a novel framework for balanced unlearning in large language models that addresses asynchronous forgetting by focusing on hard-to-unlearn data distributions. BalDRO,…
Federated learning enables healthcare facilities to train shared machine learning models without centralizing sensitive patient data, but operational challenges such as deployment, monitoring, and gov…
A new self-supervised learning framework, Tiny Vision Transformer with Pathology-Aware Prototype Distillation (TVT-PAPD), achieved a weighted F1-score of 93.02% on the TCGA LGG and GBM dataset and 90.…
Researchers have introduced a supervised learning framework that uses a quantum kernel from reduced density matrices of small subsystems to detect quantum topological phases with high accuracy, even w…
Researchers have introduced GradSkip, a new method for Vision Transformers that uses adaptive head weighting and re-evaluates residual connections to achieve a 14-times reduction in GFLOPs while impro…
Researchers have demonstrated that forcing neural networks to adopt low-dimensional representations significantly enhances their ability to generalize, according to a new study. Using information-theo…
AI shows strong promise in detecting cardiac amyloidosis (CA) through bone scintigraphy and SPECT/CT, backed by large externally validated cohorts, but lags in subtype classification and prognosis due…
New research shows depth is more important than width in neural networks for approximating analytic functions, challenging traditional thinking. The study derived new approximation rates expressed as …
The UK faces a 'triple whammy' in AI: heavy investment in AI stocks reminiscent of the dot-com bubble, slower-than-expected enterprise adoption due to change management and skills shortages, and the r…
A new framework for strong invertibility in nonlinear dynamical systems, based on bi-Lipschitz conditions, promises to improve AI model reliability in tasks such as trajectory optimization and generat…
A new adversarial attack method called GATAS exploits the latent space of text-to-speech models to achieve a 98% success rate in deceiving automatic speech recognition (ASR) systems while preserving s…
The Multi-scale Feature Enhanced Graph Neural Network (ME-GNN) achieved a relative L2 error of 0.0196 for velocity fields on the ShapeNet-Car benchmark and a normalized mean squared error of 0.0033 fo…
DynaFilter, a new technique for satellite edge data processing, enables selective inference directly in compressed domains, reducing data needed for decoding and inference by 1.6x to 7.1x for images a…
Google DeepMind CEO Demis Hassabis proposed a FINRA-style self-regulatory body for frontier AI models on July 14, calling for mandatory 30-day pre-release safety assessments. The framework would apply…
DeepSeek is in preliminary talks to raise new funds at a roughly $71 billion valuation, a $19 billion increase from its $52 billion valuation set six weeks ago in May, according to the Financial Times…
Thomson Reuters told staff on July 13 it will cut up to 500 engineering roles while hiring more than 250 AI-native engineers globally, a swap of roughly two traditional engineers out for every AI spec…
New York Governor Kathy Hochul signed an executive order on July 14 making New York the first US state to pause new permits for data centers drawing more than 50 megawatts of power for up to a year, g…
OpenAI-backed Chai Discovery closed a $400 million funding round at a $3.8 billion valuation on July 14, tripling its price tag in seven months. The Index Ventures-led round positions the company as A…
SPARC-Net, a novel neural network architecture, claims to overcome the limitations of Physics-Informed Neural Networks (PINNs) in solving stiff and shock-dominated partial differential equations, achi…
A new unified multi-dimensional priority-constraint framework for physics-informed neural networks (PINNs) prioritizes training regions based on natural physical information flow, using negative-expon…