Cooperative multi-agent systems often face tasks that require coordinated actions under uncertainty. While multi-armed bandit (MAB) problems provide a powerful framework for decentralized learning, most prior work assumes individually attainable rewards. We address the challenging setting where rewards are threshold-activated: an arm yields a payoff only when a minimum number of agents pull it simultaneously, with this threshold unknown in advance. Complicating matters further, some arms are decoys - requiring coordination to activate but yielding no reward - introducing a new challenge of wasted joint exploration. We introduce Threshold-Coop-UCB (T-Coop-UCB), a decentralized algorithm that enables agents to jointly learn activation thresholds and reward distributions, forming effective coalitions without centralized control. Empirical results show that T-Coop-UCB consistently outperforms baseline methods in cumulative reward, regret, and coordination metrics, achieving near-Oracle performance. Our findings underscore the importance of joint threshold learning and decoy avoidance for scalable, decentralized cooperation in complex multi-agent

Power grid operation is becoming increasingly complex due to the rising integration of renewable energy sources and the need for more adaptive control strategies. Reinforcement Learning (RL) has emerged as a promising approach to power network control (PNC), offering the potential to enhance decision-making in dynamic and uncertain environments. The Learning To Run a Power Network (L2RPN) competitions have played a key role in accelerating research by providing standardized benchmarks and problem formulations, leading to rapid advancements in RL-based methods. This survey provides a comprehensive and structured overview of RL applications for power grid topology optimization, categorizing existing techniques, highlighting key design choices, and identifying gaps in current research. Additionally, we present a comparative numerical study evaluating the impact of commonly applied RL-based methods, offering insights into their practical effectiveness. By consolidating existing research and outlining open challenges, this survey aims to provide a foundation for future advancements in RL-driven power grid optimization.

Event-based vision sensors produce an asynchronous stream of events which are triggered when the pixel intensity variation exceeds a predefined threshold. Such sensors offer significant advantages, including reduced data redundancy, micro-second temporal resolution, and low power consumption, making them valuable for applications in robotics and computer vision. In this work, we consider the problem of video reconstruction from events, and propose an approach for dynamic feedback control of activation thresholds, in which a controller network analyzes the past emitted events and predicts the optimal distribution of activation thresholds for the following time segment. Additionally, we allow a user-defined target peak-event-rate for which the control network is conditioned and optimized to predict per-column activation thresholds that would eventually produce the best possible video reconstruction. The proposed OnTheFly control scheme is data-driven and trained in an end-to-end fashion using probabilistic relaxation of the discrete event representation. We demonstrate that our approach outperforms both fixed and randomly-varying threshold schemes by 6-12% in terms of LPIPS perceptual image dissimilarity metric, and by 49% in terms of event rate, achieving superior reconstruction quality while enabling a fine-tuned balance between performance accuracy and the event rate. Additionally, we show that sampling strategies provided by our OnThe-Fly control are interpretable and reflect the characteristics of the scene. Our results, derived from a physically-accurate simulator, underline the promise of the proposed methodology in enhancing the utility of event cameras for image reconstruction and other downstream tasks, paving the way for hardware implementation of dynamic feedback EVS control in silicon.

Polynomial neural networks are important in applications and theoretical machine learning. The function spaces and dimensions of neurovarieties for deep linear networks have been studied, and new developments in the polynomial neural network setting have appeared. In particular, results on the choice of the activation degree and the dimension of the neurovariety have improved our understanding of the optimization process of these neural networks and the ability of shallow and deep neural networks to replicate target functions [21, 27]. These theoretical results possess relevant implications. For appropriate datasets, polynomial activation functions can reduce model complexity and computational costs by introducing higher-order interactions between inputs, making it possible to model non-linear phenomena more efficiently. Moreover, polynomial neural networks have been found to perform well in practice in high-impact fields such as healthcare and finance.

Activation sparsity refers to the existence of considerable weakly-contributed elements among activation outputs. As a prevalent property of the models using the ReLU activation function, activation sparsity has been proven a promising paradigm to boost model inference efficiency. Nevertheless, most large language models (LLMs) adopt activation functions without intrinsic activation sparsity (e.g., GELU and Swish). Some recent efforts have explored introducing ReLU or its variants as the substitutive activation function to help LLMs achieve activation sparsity and inference acceleration, but few can simultaneously obtain high sparsity and comparable model performance. This paper introduces a simple and effective sparsification method named "ProSparse" to push LLMs for higher activation sparsity while maintaining comparable performance. Specifically, after substituting the activation function of LLMs with ReLU, ProSparse adopts progressive sparsity regularization with a factor smoothly increasing along the multi-stage sine curves. This can enhance activation sparsity and mitigate performance degradation by avoiding radical shifts in activation distributions. With ProSparse, we obtain high sparsity of 89.32% for LLaMA2-7B, 88.80% for LLaMA2-13B, and 87.89% for end-size MiniCPM-1B, respectively, achieving comparable performance to their original Swish-activated versions. These present the most sparsely activated models among open-source LLaMA versions and competitive end-size models, considerably surpassing ReluLLaMA-7B (66.98%) and ReluLLaMA-13B (71.56%). Our inference acceleration experiments further demonstrate the significant practical acceleration potential of LLMs with higher activation sparsity, obtaining up to 4.52$\times$ inference speedup.

Understanding how high-level concepts are represented within artificial neural networks is a fundamental challenge in the field of artificial intelligence. While existing literature in explainable AI emphasizes the importance of labeling neurons with concepts to understand their functioning, they mostly focus on identifying what stimulus activates a neuron in most cases; this corresponds to the notion of recall in information retrieval. We argue that this is only the first-part of a two-part job; it is imperative to also investigate neuron responses to other stimuli, i.e., their precision. We call this the neuron label's error margin.

One of the most frequent and severe aftermaths of a stroke is the loss of upper limb functionality. Therapy started in the sub-acute phase proved more effective, mainly when the patient participates actively. Recently, a novel set of rehabilitation and support robotic devices, known as supernumerary robotic limbs, have been introduced. This work investigates how a surface electromyography (sEMG) based control strategy would improve their usability in rehabilitation, limited so far by input interfaces requiring to subjects some level of residual mobility. After briefly introducing the phenomena hindering post-stroke sEMG and its use to control robotic hands, we describe a framework to acquire and interpret muscle signals of the forearm extensors. We applied it to drive a supernumerary robotic limb, the SoftHand-X, to provide Task-Specific Training (TST) in patients with sub-acute stroke. We propose and describe two algorithms to control the opening and closing of the robotic hand, with different levels of user agency and therapist control. We experimentally tested the feasibility of the proposed approach on four patients, followed by a therapist, to check their ability to operate the hand. The promising preliminary results indicate sEMG-based control as a viable solution to extend TST to sub-acute post-stroke patients.

Deep Reinforcement Learning agents often suffer from catastrophic forgetting, forgetting previously found solutions in parts of the input space when training on new data. Replay Memories are a common solution to the problem, decorrelating and shuffling old and new training samples. They naively store state transitions as they come in, without regard for redundancy. We introduce a novel cognitive-inspired replay memory approach based on the Grow-When-Required (GWR) self-organizing network, which resembles a map-based mental model of the world. Our approach organizes stored transitions into a concise environment-model-like network of state-nodes and transition-edges, merging similar samples to reduce the memory size and increase pair-wise distance among samples, which increases the relevancy of each sample. Overall, our paper shows that map-based experience replay allows for significant memory reduction with only small performance decreases.

A current research trend in neurocomputing involves the design of novel artificial neural networks incorporating the concept of time into their operating model. In this paper, a novel architecture that employs stigmergy is proposed. Computational stigmergy is used to dynamically increase (or decrease) the strength of a connection, or the activation level, of an artificial neuron when stimulated (or released). This study lays down a basic framework for the derivation of a stigmergic NN with a related training algorithm. To show its potential, some pilot experiments have been reported. The XOR problem is solved by using only one single stigmergic neuron with one input and one output. A static NN, a stigmergic NN, a recurrent NN and a long short-term memory NN have been trained to solve the MNIST digits recognition benchmark.

I'm a developer from the UK who loves technology and business. Here you'll find articles and tutorials about things that interest me.