Bo Li

This study investigates how retrieval-augmented models improve reliability in language generation.

A new method enhances climate model comparison with real-world observations.

New strategies allow Trojan attacks to bypass detection methods effectively.

A new method generates fake defective samples to improve anomaly detection in manufacturing.

This study investigates how elliptical shapes of coronal loops affect their movement.

Exploring new privacy concerns in the use of diffusion models.

Exploring the balance between model compression and trustworthiness in AI.

A novel approach to enhance moderation methods for large language models.

A new framework for training models without sharing sensitive data.

Introducing Mixture of Low-Rank Experts for improved multi-task learning performance.

ALERT benchmark assesses safety risks in language models to improve their responses.

A new framework enhances prediction confidence through learning and reasoning.

This article covers kink oscillations and their significance in solar corona studies.

Examining client contributions and vulnerabilities in federated learning systems.

A new framework enhances prediction accuracy and data privacy in federated learning.

A framework to enhance safety in LLM agents across various applications.

A new method tackles hidden threats in large language models.

A look at AI risk categories and the need for unified policies.

A new system optimizes training for large AI models by reducing communication costs.

A new model improves safety monitoring for large language models against harmful content.

A study on improving trap placement for better WNV monitoring.

A new benchmark aims to assess AI safety risks effectively.

A new approach improves decision-making in AI through causal relationships.

Analyzing the importance and difficulties of assessing multimodal AI models.

Exploring the unique behaviors of chirality in silver isotopes.

A new method improves the training process for complex AI models.

A new method improves tamper resistance in open-weight language models.

A new approach to assess LLMs with diverse evaluation sets.

This study examines envy-free resource distribution in graph settings.

Examining how a volcanic eruption shaped climate patterns globally.

This article discusses the safety and security issues in multimodal AI systems.

A look into the fascinating movements in the Sun's outer layer.

This dataset enhances question generation from educational diagrams for better student learning.

Scientists study how waves behave in messy environments, revealing surprising patterns and potential benefits.

Discover how Composition of Experts transforms AI effectiveness and efficiency.

New method enables backdoor attacks without clean data or model changes.

Discover how CALM transforms image evaluation with AI-driven insights.

Learn how intention knowledge graphs enhance online shopping experiences.

M-ALERT tests language models for safety across five languages.