AI/ML arXiv cs.AI

A Framework of User Experience Principles for Human-AI Agent Interaction in the Workplace

A research paper proposing a design framework of eight UX principles for human-AI agent interactions in the workplace.

AI/ML arXiv cs.AI

G-MAD: A Game-Based Data Generation Framework for Multi-View RGB-T Aerial Object Detection

G-MAD is an open-source framework using Arma3 to generate synthetic multi-view RGB-T data for aerial object detection.

AI/ML arXiv cs.AI

When Does Knowledge Distillation Hurt? Reliability-Aware Distillation for Low-Resource Language Summarization

Research introducing reliability-aware distillation methods (CHAD and EWAD+CPDP) to improve low-resource language summarization.

Cybersecurity arXiv cs.AI

HijackKV: New Threat in Position-Independent KV Cache Reuse

Introduction of HijackKV, an attack framework exploiting vulnerabilities in position-independent KV cache reuse in LLMs.

AI/ML arXiv cs.AI

When Shippers Become Algorithms: Candidate Exposure, Information Design, and the Concentration of LLM-Mediated Freight Markets

A study on how LLM-mediated freight markets lead to carrier concentration and the impact of platform information design.

AI/ML arXiv cs.AI

Time Series Network Utilization KPI Forecasting Using Advanced AI/ML Models

Evaluation of various AI/ML models for forecasting network bandwidth utilization to improve resource provisioning.

Tech Business/VC Hacker News

EU fines Google €890M for competition breaches over search and apps

The EU has fined Google €890 million for violating competition rules regarding search and app store practices.

Tech Business/VC The Verge

Google hit with $1 billion fine for breaking EU antitrust rules

Google faces a $1 billion fine from the EU for preferential treatment of its own services in search and restricting alternative payment options in the Play Store.

AI/ML arXiv cs.AI

Sentence Splitter: Uncovering Latent Factual Structure for Self-Supervised Learning

Researchers introduce Sentence Splitter, a self-supervised framework using a T5 encoder-decoder to uncover latent factual structures in natural language for better NLP performance.

AI/ML arXiv cs.AI

Auto-Fill: Learning to Predict Missing Values Accurately with Specialist Language Models

Auto-Fill uses an ensemble of three specialist small language models to accurately predict missing values in tabular data at a fraction of the cost of frontier models.

AI/ML arXiv cs.AI

Overview of FinMMEval 2026 Task 1: Multilingual Financial Multiple-Choice Question Answering

FinMMEval 2026 Task 1 provides a benchmark for evaluating multilingual financial multiple-choice question answering across English, Chinese, Arabic, and Hindi.

AI/ML arXiv cs.AI

Memory-Augmented Multimodal Large Language Models for Small Object Understanding in Streaming Aerial Videos

The SkyAnchor MLLM and DroneEyes dataset are introduced to improve small object understanding in streaming aerial videos for UAVs.

AI/ML arXiv cs.AI

PRISM-DR: Per-lesion Retinal Inference with Specialist Models for Diabetic Retinopathy

PRISM-DR proposes a lesion-specific pipeline using multiple YOLO detectors to improve the detection of diverse diabetic retinopathy lesions in retinal images.

AI/ML arXiv cs.AI

Overview of FinMMEval 2026 Task 2: Multilingual Financial Short-Answer Question Answering

FinMMEval 2026 Task 2 evaluates multilingual financial short-answer question answering based on reports in multiple languages.

Cybersecurity arXiv cs.AI

Defense Against LLM Backdoors using Critical Neuron Isolation Pruning

DeCNIP is a new defense mechanism that identifies and prunes critical neurons to neutralize LLM backdoors while maintaining overall model utility.

AI/ML arXiv cs.AI

OSVE: One Step Video Editing with One Step Diffusion Models

OSVE is a one-step diffusion framework for high-quality video editing that is significantly faster than traditional multi-step methods.

Other Hacker News

Frequently Asked Questions on Expertise

A Hacker News discussion regarding the conceptual and practical aspects of expertise.

Cybersecurity The Verge

Google now lets you sign in to your account using a selfie video

Google introduces a selfie video verification method for account recovery to help users regain access to locked accounts.

AI/ML arXiv cs.AI

The World Model Remembers, the Actor Forgets: Dream Rehearsal for Continual Model-Based RL

Researchers propose 'dream rehearsal' using supervised self-imitation on a world model's dreams to prevent catastrophic forgetting in Model-Based RL agents.

AI/ML arXiv cs.AI

Learning the Arabic Dialect Continuum as a Continuous Space: A Regression Approach to Speaker Origin Prediction

A new regression-based approach to Arabic dialect geolocation models dialectal variation as continuous geographic coordinates using a hierarchical neural architecture.