IEEE

GIGAS: Adversarial Attacks on Visual Question Answering with Multi-modal Generative Models

Abstract: VQA models, which answer questions about images by combining both visual and textual information, have been proven susceptible to adversarial attacks. These attacks introduce subtle ...
GitHub

RoboTrust: Evaluating the Interaction Trustworthiness of Multi-modal Large Language Models in Embodied Agents

Formalizes embodied trust through 5 dimensions and 12 fine-grained tasks, covering factual perception, ethical safety, bias control, environmental robustness, and privacy protection. [2025/10/6] 🤖 We ...
9to5Mac

Perplexity Computer adding ability to split tasks between local and cloud models

Perplexity has announced a major new feature coming soon to Perplexity Computer: the ability to split tasks between local and cloud models. Perplexity Computer is the company’s agentic system for ...
IEEE

Large Models Empowered End-to-End AI for Split Learning in Native AI Future Networks

Abstract: The emergence of sixth-generation mobile communication networks (6G) marks a paradigm shift toward native artificial intelligence (AI) architectures, where AI is deeply embedded in the ...