Latest Articles for Safety In Technology

MLGuard offers a framework for safe and reliable machine learning applications.

Adv3D introduces realistic 3D adversarial examples for self-driving car systems.

A new method for safe robot navigation considering physical risks in complex environments.

Introducing a benchmark to verify neural networks coded in plain C.

A new method enhances AV models' adaptability to camera viewpoint changes.

Examining the role of testing tools in ensuring safe image generation.

Discussing safety measures for machine learning systems in critical areas.

Exploring the need for safety and trust in AI systems.

Introducing a framework that prioritizes safety in reinforcement learning.

A look into the importance of model verification for safety in AI systems.

New methods improve predictions for spent nuclear fuel management.

Explore how Data-Driven Safety Filters maintain safety in learning-based systems.

A new approach to monitor neural networks during operation to ensure reliability.

New methods improve safe navigation for autonomous systems through innovative control strategies.

New methods improve the safety verification of Bayesian Neural Networks against attacks.

New methods improve decision-making in AI while ensuring safety and efficiency.

This study assesses the safety of DNNs in handling unfamiliar driving data.

A look at control barrier functions for safe operation in technology.

A look into safe reinforcement learning techniques and their real-world applications.

A new method enhances safety for robots navigating complex environments.

A new method enhances how robots interpret user instructions safely.

Exploring safety measures to tackle overconfidence in autonomous vehicle AI systems.

A new framework for improving safety and performance in hybrid robotic systems.

An overview of distributed systems, synchronization, and safety methods.

A community-led initiative to identify harmful prompts in T2I models.

A new approach to ensure safety in offline reinforcement learning.

A new method aims to enhance large language models' safety and usefulness.

A new framework enhances the safety of UGVs in complex environments.

A study on Shielded Deep Reinforcement Learning for safe spacecraft autonomy.

A new method enhances safety features in multimodal AI systems without extensive training.

A guide on combining AI with robots for safer and efficient operations.

New methods enhance safety in reinforcement learning while optimizing performance in constrained environments.

DeepKnowledge method improves the reliability of DNNs in critical applications.

A new method enhances OOD detection by focusing on gradient information.

Research focuses on improving neural network verification with minimal NAP specifications.

DEXTER improves safety for AI by enhancing out-of-distribution detection.

A study comparing the safety performance of popular language models.

Combining OOD detection and Conformal Prediction enhances model reliability.

Learn how breaking down complex tasks helps robots navigate effectively.

A new dataset evaluates how language models handle harmful content across cultures.