Frontiers

Brain-Inspired and neurally grounded algorithms in learning and control of advanced robots

In recent years, the field of robotics has undergone significant transformation, driven increasingly by advances in brain-inspired and neurally grounded ...

New 'Markovian Thinking' technique unlocks a path to million-token AI reasoning

The 'Delethink' environment trains LLMs to reason in fixed-size chunks, breaking the quadratic scaling problem that has made long-chain-of-thought tasks prohibitively expensive.
Communications of the ACM

Shields for Safe Reinforcement Learning

Evaluating the advantages and potential drawbacks of shielding as a method for safe RL. Bettina Könighofer is an assistant ...
Frontiers

Prediction of long-term adherence to direct oral anti-coagulants in patients with atrial fibrillation using first-order Markov models

Direct Oral Anti-Coagulants (DOACs) are the primary treatment for the long-term prevention of stroke in patients with atrial fibrillation. Strict adherence to DOAC therapy is crucial and must be ...
IEEE

Distributional Black-Box Model Inversion Attack With Multi-Agent Reinforcement Learning

Abstract: Model Inversion (MI) attacks based on Generative Adversarial Networks (GAN) aim to recover private training data from complex deep learning models by searching codes in the latent space.
IEEE

Risk-Aware Modeling of Control Loss Using Reinforcement Learning and Partially Observable Markov Decision Process Models

Abstract: This article presents an approach for identifying high-risk control loss situations in digital environments by combining reinforcement learning methods with decision-making models under ...