Latest Articles for Robustness

Research investigates how simple models exhibit complex behaviors and agency.

Examining the dangers of integrating language models in robotic systems.

Researchers investigate non-Abelian braiding operations in quantum states for improved computing.

This article discusses methods to certify explanations in machine learning models.

A system to check fairness in machine learning while protecting model privacy.

A new method enhances deep learning models' strength and accuracy.

Leveraging CLIP's visual and text components improves deepfake detection methods.

Research focuses on smooth safety filters for improved autonomous vehicle performance.

A new watermarking method safeguards customized language models from unauthorized use.

A study on the role of Degenerate Knowledge Neurons in improving language model performance.

Discover how topological quantum devices enhance quantum technology applications.

A new collective approach enhances counterfactual explanations in decision-making processes.

A look at the ProTIP framework for assessing AI image generation models.

This article reviews issues with AI watermarking and suggests improvements for security.

Discover how machine learning enhances anomaly detection in particle physics research.

Examining the impact of miscalibration on NLP models' resilience to adversarial attacks.

This article discusses the key issues in the growth of generative AI.

Researchers combine trajectory optimization and reinforcement learning to enhance legged robots.

BNQN offers a smoother and more robust alternative for finding roots in mathematics.

Exploring methods to improve the reliability of AI decision-making explanations.

A new approach improves machine learning accuracy while ensuring data privacy.

A new method offers reliable estimation for diverse data types.

Auxiliary Fourier-basis Augmentation enhances image robustness in computer vision models.

A new method enhances resilience of models to adversarial examples through text prompt adjustment.

A new method for consistent counterfactual explanations despite model changes.

The EHT captures the first image of a black hole's shadow, confirming key theories.

WaterMax improves watermarking in AI-generated text, ensuring quality and traceability.

Understanding model robustness is key for real-world applications in various fields.

A new method enhances deep learning model resilience against attacks using peer models.

Exploring the need for clear AI specifications to tackle bias and transparency.

Investigating model compression methods to improve efficiency and defenses against attacks.

AVIBench tests LVLMs to ensure they withstand adversarial visual instructions.

Improving robustness against adversarial attacks in vision-language models.

This article reviews the robustness of CLIP in various challenges.

This article reviews the strengths and weaknesses of the VMamba model.

A new method for assessing image quality without human ratings shows promising results.

A look at how machines analyze and interpret visual data.

This article presents a method to reduce bias in AI predictions.

Tropical neural networks enhance resilience against adversarial attacks in machine learning.

A new framework improves video captioning accuracy despite missing inputs.