Beyond Curiosity: Rewarding Agents for Changing Their Minds
![The system introduces a method for guiding policy learning through intrinsic rewards derived from both the consistency of strategic embeddings across state transitions-quantified as Strategy Stability-and the magnitude of prediction error coupled with shifts in strategy-captured by Strategy Surprise [latex] r_{int} [/latex].](https://arxiv.org/html/2601.10349v1/x1.png)
A new reinforcement learning framework, Strategy-aware Surprise, encourages more effective exploration by focusing on shifts in an agent’s behavioral approach, not just novel states.
Science
Beyond Bar and Line: A New Dataset for Smarter Chart Understanding
Researchers introduce ChartComplete, a comprehensive resource designed to push the boundaries of AI’s ability to interpret a wider range of chart types.
Fooling the Face: Stealthy Attacks and Forensic Defenses
Researchers are exploring how subtly crafted image patches, generated by advanced AI models, can bypass facial recognition systems, and the techniques to detect these deceptive alterations.
Decoding Data Trails: AI Analyzes Privacy Policies for Connected Cars
![The architecture anticipates eventual failure, embracing a layered defense-[latex]LADFA[/latex]-designed not to prevent cascading errors, but to contain their inevitable spread.](https://arxiv.org/html/2601.10413v1/x1.png)
A new framework uses the power of large language models to automatically map how personal data flows within the complex ecosystems of connected vehicle mobile applications.
Decoding Heart Rhythms: A New Approach to Atrial Fibrillation Detection
Researchers have developed a deep learning framework that combines time and frequency analysis of ECG signals for more accurate and reliable detection of atrial fibrillation.
Learning by Reading: Agents Trained on Tool-Use Stories
Researchers are demonstrating a new method for training autonomous agents to use tools by extracting procedural knowledge directly from natural language text.
Spotting the Errors: A Smarter Approach to Data Cleaning
New research introduces a method for automatically identifying mislabeled data points, boosting the accuracy of machine learning models.
Seeing Isn’t Believing: How SHAP Values Amplify Vision Model Attacks
New research reveals that leveraging SHAP values in adversarial attacks significantly increases the success rate of misclassifying images in computer vision systems.
Seeing is Believing: AI Sheds Light on Pediatric Pneumonia
New research demonstrates how deep learning, coupled with explainable AI techniques, can accurately identify pneumonia in children’s chest X-rays and provide clinicians with crucial insights into its decision-making process.
Bridging the Digital Gap for Seniors with Smarter AI
New research explores how understanding the unique communication styles of older adults can lead to more effective and accessible AI-powered technology support.