Research & Innovation

Global: Deep Surrogate Framework Predicts Quality of Error-Bounded Lossy CompressionNew Framework Targets Compression Quality PredictionResearchers have introduced a deep‑surrogate framework called DeepCQ that predicts the quality of error‑bounded lossy compression…

Global: New Equivariance Framework Links Symmetry to Curvature in Deep LearningResearchers announced a novel analytical toolbox in a December 2025 arXiv preprint that extends Noether‑type symmetry analyses to both first‑…

Global: Backdoor Threats Exposed in Retrieval-Augmented Code Generation SystemsResearchers have uncovered a practical backdoor vulnerability affecting the retriever component of retrieval-augmented code generation (RACG) frameworks, according to a paper posted…

Global: New ‘Verifiable Passkey’ Standard Addresses Storage Limits and Privacy ConcernsA recently posted arXiv paper introduces a novel “Verifiable Passkey” standard designed to let users employ passkeys issued by a…

Global: Researchers Demonstrate Fully Analogue Neural Networks Using NDR DevicesA team of researchers has introduced KANalogue, a fully analogue implementation of Kolmogorov‑Arnold Networks that leverages negative‑differential‑resistance (NDR) devices to perform…

Global: AI-Driven System ‘Quicker’ Aims to Streamline Clinical Evidence SynthesisIn May 2025, researchers posted a study on arXiv describing the introduction of Quicker, an evidence‑based clinical decision support system powered…

Global: New Zeroth-Order Framework Enhances Privacy and Efficiency in Vertical Federated LearningBreakthrough in Private Model TrainingResearchers have introduced DPZV, a communication‑efficient, differentially private zeroth‑order vertical federated learning (VFL) framework that…

Global: Magnitude-Driven LoRA Initialization Reduces Overhead While Matching Spectral PerformanceIn a July 2025 preprint posted to arXiv, researchers introduced LoRAM, a magnitude‑driven initialization scheme for Low‑Rank Adaptation (LoRA) that aims…

Global: Study Quantifies Security Benefits of Rotating QKD-Derived Keys for Block CiphersResearchers have introduced a precise calculation model that determines how often quantum‑generated keys should be rotated when used with…

Global: New Framework Improves Uncertainty Estimation in Social Bot DetectionResearchers have introduced a novel Uncertainty Estimation for Social Bot Detection (UESBD) framework in a paper posted to arXiv in March…