Many embedded applications have strict energy, memory, and time constraints, making neural network (NN) inference particularly challenging. Recently, a novel NN architecture, called Fast Feedforward Networks (FFFs), has been proposed to achieve inference with extremely lightweight computational demands and minimal latency. Yet, compared to feedforward networks with similar sizes, FFFs still lag behind in terms of performance, indicating that they do not utilize all of their parameters effectively. In this article, we explore a possible reason for this performance gap: the uncertainty in how samples are assigned to the network’s leaves. We attempt to overcome this challenge by making FFFs’ training inference-aware, hence introducing Inference-Aware Fast Feedforward Networks (IAFFFs). We imitate FFFs’ inference during training by using a step activation function alongside the traditional sigmoid activation function. We test different aware scheduling methods, which we dub “awareness scheduler”, to adjust the balance between the two activation functions during training, and examine how different schedules impact the model’s performance. Additionally, we employ leaf-weight virtualization with inference-aware retraining to compress our models so they can fit onto edge devices. We further employ an iterative compression approach to find an optimal awareness scheduler for compression to minimize performance drop due to compression. We experiment with different model sizes on various microcontrollers (MCUs) with different memory constraints to observe the latency and energy consumption introduced by the compression algorithm.

In the past few years, Federated Learning (FL) has emerged as an effective approach for training Neural Networks (NNs) over a computing network while preserving data privacy. Most existing FL approaches require defining a priori (1) a predefined structure for all the NNs running on the clients and (2) an explicit aggregation procedure. These can be limiting factors in cases where predefining such algorithmic details is difficult. Recently, NEvoFed was proposed, an FL method that leverages Neuroevolution running on the clients, in which the NN structures are heterogeneous and the aggregation is implicitly accomplished on the client side. Here, we propose MFC-NEvoFed, a novel approach to FL that does not require learning models, i.e., neural network parameters, to be distributed over the networks, thus taking a step toward security improvement. The only information exchanged in client/server communication is the performance of each model on local data, allowing the emergence of optimal NN architectures without needing any kind of model aggregation. Another appealing feature of our framework is that it can be used with any Machine Learning algorithm provided that, during the learning phase, the model updates do not depend on the input data. To assess the validity of MFC-NEvoFed, we test it on four datasets, showing that very compact NNs can be obtained without drops in performance compared to canonical FL. Finally, such compact structures allow for a step toward explainability, which is highly desirable in domains such as digital health, from which the tested datasets come.

Continuous control tasks often involve high-dimensional, dynamic, and non-linear environments. State-of-the-art performance in these tasks is achieved through complex closed-box policies that are effective, but suffer from an inherent opacity. Interpretable policies, while generally underperforming compared to their closed-box counterparts, advantageously facilitate transparent decision-making within automated systems. Hence, their usage is often essential for diagnosing and mitigating errors, supporting ethical and legal accountability, and fostering trust among stakeholders. In this paper, we propose SMoSE, a novel method to train sparsely activated interpretable controllers, based on a top-1 Mixture-of-Experts architecture. SMoSE combines a set of interpretable decision-makers, trained to be experts in different basic skills, and an interpretable router that assigns tasks among the experts. The training is carried out via state-of-the-art Reinforcement Learning algorithms, exploiting load-balancing techniques to ensure fair expert usage. We then distill decision trees from the weights of the router, significantly improving the ease of interpretation. We evaluate SMoSE on six benchmark environments from MuJoCo: our method outperforms recent interpretable baselines and narrows the gap with non-interpretable state-of-the-art algorithms.

Fall resilience remains a significant challenge for robots capable of locomotion. In the case of humanoid robots, arms can be utilized to mitigate impact damage, thereby improving resilience to falls. Although previous work has primarily focused on damage reduction strategies for legged robots, the question of minimizing fall damage in humanoid ballbots remains largely unexplored. In this paper, we introduce a novel multi-policy approach, combining clustering on the initial pose and Reinforcement Learning, to reduce damage resulting from a fall in a commercial humanoid ballbot. We conduct simulations to compare our method against three baselines: (1) a curriculum learning policy, (2) a single-policy approach enhanced by clustering information, and (3) a standard single-policy baseline. Experimental results demonstrate that our approach achieves higher success rates and greater damage reduction compared to existing alternatives.

An old-school recipe for training a classifier is to (i) learn a good feature extractor and (ii) optimize a linear layer atop. When only a handful of samples are available per category, as in Few-Shot Adaptation (FSA), data are insufficient to fit a large number of parameters, rendering the above impractical. This is especially true with large pre-trained Vision-Language Models (VLMs), which motivated successful research at the intersection of Parameter-Efficient Fine-tuning (PEFT) and FSA. In this work, we start by analyzing the learning dynamics of PEFT techniques when trained on few-shot data from only a subset of categories, referred to as the "base" classes. We show that such dynamics naturally splits into two distinct phases: (i) task-level feature extraction and (ii) specialization to the available concepts. To accommodate this dynamic, we then depart from prompt- or adapter-based methods and tackle FSA differently. Specifically, given a fixed computational budget, we split it to (i) learn a task-specific feature extractor via PEFT and (ii) train a linear classifier on top. We call this scheme Two-Stage Few-Shot Adaptation (2SFS). Differently from established methods, our scheme enables a novel form of selective inference at a category level, i.e., at test time, only novel categories are embedded by the adapted text encoder, while embeddings of base categories are available within the classifier. Results with fixed hyperparameters across two settings, three backbones, and eleven datasets, show that 2SFS matches or surpasses the state-of-the-art, while established methods degrade significantly across settings. .

Task failures in decentralized Internet of Things (IoT)-edge computing environments not only lead to inefficiencies, increased latency, and resource wastage but can also introduce system instability and cause application malfunctions. These failures may arise due to network disruptions, resource constraints, or inefficient task scheduling, ultimately affecting the overall reliability and performance of IoT-edge systems. This study presents a novel Long Short-Term Memory (LSTM)-based Federated Learning (FL) framework for proactive task failure prediction, ensuring adaptive scheduling and efficient resource utilization. Unlike existing conventional methods, our approach personalizes failure prediction per device, addressing heterogeneous execution characteristics while preserving data privacy. By integrating LSTM with FL, we improve the failure detection accuracy and reduce unnecessary task executions. We first trained all models using Federated Learning (FL) and then conducted a comparative analysis of Convolutional Neural Networks (CNN), Gated Recurrent Units (GRU), and LSTM. Our findings show that LSTM achieves the highest accuracy and F1 score, while CNN excels in recall and energy efficiency. These insights validate the effectiveness of our FL-based failure prediction framework and highlight the advantages of model personalization for dynamic decentralized IoT-edge environments.

Human Activity Recognition (HAR) has profound applications in domains such as healthcare, wearable devices, and smart environments, where continuous monitoring is essential. However, traditional centralized learning approaches raise privacy concerns and are infeasible for resource-constrained devices due to high communication costs. Federated Learning (FL) offers a privacy-preserving solution by training models locally and aggregating updates, but it still encounters considerable communication overhead. This paper proposes an optimized FL framework that integrates model compression through Knowledge Distillation (KD) to reduce communication costs while preserving model performance. Using a Teacher-Student model architecture, the framework enables the deployment of highly compressed Student models without compromising accuracy or other performance metrics. Experimental evaluations on a smartphone-based HAR dataset show that the proposed framework achieves comparable accuracy, precision, and recall w.r.t. the original (uncompressed) models, with up to 75 % reduction in model size (and, consequently, communication cost). This approach demonstrates scalability and feasibility for real-world HAR applications on resource-constrained devices, also laying the groundwork for efficient, privacy-preserving distributed learning in heterogeneous computational environments.

Modern agriculture has enabled food production for nearly eight billion people, yet plant diseases and climate change continue to threaten food security. Beans, a key nutritional crop worldwide, are vulnerable to diseases such as bean rust and angular leaf spot, which significantly reduce yield. Early detection is critical for effective treatment. While existing methods leverage cloud-based Deep Learning (DL) models for plant disease classification, they often require substantial computational resources. In this work, we propose an edge-based solution by deploying a quantized MobileNetV2 model on a resourceconstrained embedded device. We compare the performance of the full-precision (Float 32-bit) model and its quantized (Int 8-bit) counterpart deployed on a microcontroller unit (MCU), targeting Internet of Things (IoT) scenarios. Our study analyzes model accuracy, inference speed, and memory utilization, focusing on peak RAM and flash memory requirements. The results show that quantization significantly reduces memory footprint and inference time while maintaining competitive classification accuracy. These findings highlight the potential of quantization to enable efficient, sustainable, and deployable TinyML models for plant disease detection at the edge.

Accurate energy forecasting and occupancy detection are critical for energy management and occupant comfort in smart buildings. Centralized models, for example, based on LSTM and GRU, perform well with homogeneous datasets but face challenges in distributed IoT settings due to privacy concerns and sensor data heterogeneity (e.g., light, temperature, humidity,$\text{CO}_{2}$). In this paper, we propose Federated Forecasting (FedFor), a privacy-preserving framework integrating Federated Learning (FL), Multi-Task Learning (MTL), and dual attention mechanisms. FedFor simultaneously enhances energy forecasting and occupancy detection by leveraging task-specific and temporal patterns while keeping data localized. We evaluated our approach on two datasets, namely ThingSpeak and Occupancy Detection Data (ODD). Results show that FedFor outperforms the baselines, achieving 99.11% accuracy for occupancy detection and reducing Mean Absolute Error (MAE) to 0.0097 for forecasting. Compared to state-of-the-art, our FedFor method is effective in addressing privacy concerns and data heterogeneity.

Multi-cellular organisms typically originate from a single cell, the zygote, that then develops into a multitude of structurally and functionally specialized cells. The potential of generating all the specialized cells that make up an organism is referred to as cellular ‘‘totipotency’’, a concept introduced by the German plant physiologist Haberlandt in the early 1900s. In an attempt to reproduce this mechanism in synthetic organisms, we present a model based on a kind of modular robot called Voxel-based Soft Robot (VSR), where both the body, i.e., the arrangement of voxels, and the brain, i.e., the Artificial Neural Network (ANN) controlling each module, are subject to an evolutionary process aimed at optimizing the locomotion capabilities of the robot. In an analogy between totipotent cells and totipotent ANN-controlled modules, we then include in our model an additional level of adaptation provided by Hebbian learning, which allows the ANNs to adapt their weights during the execution of the locomotion task. Our in silico experiments reveal two main findings. Firstly, we confirm the common intuition that Hebbian plasticity effectively allows better performance and adaptation. Secondly and more importantly, we verify for the first time that the performance improvements yielded by plasticity are in essence due to a form of specialization at the level of single modules (and their associated ANNs): thanks to plasticity, modules specialize to react in different ways to the same set of stimuli, i.e., they become functionally and behaviorally different even though their ANNs are initialized in the same way. This mechanism, which can be seen as a form of totipotency at the level of ANNs, can have, in our view, profound implications in various areas of Artificial Intelligence (AI) and applications thereof, such as modular robotics and multi-agent systems.

Vehicular Edge Computing (VEC) is a key enabler of real-time intelligence in next-generation transportation systems. However, conventional Federated Learning (FL) in VEC typically depends on static edge-server aggregation, resulting in high communication overhead, increased latency, and poor responsiveness under dynamic mobility. To overcome these challenges, we propose Proximity-Aware Federated Learning (PA-FL), a decentralized framework that integrates vehicle-to-vehicle (V2V) collaboration and edge-assisted synchronization to enhance learning efficiency, scalability, and robustness. PA-FL introduces three core innovations: (i) Collaborative Local Aggregation, where vehicles perform proximity-based model fusion before forwarding updates to the edge, reducing uplink traffic and accelerating convergence; (ii) Adaptive Neighbor Selection, which dynamically filters peers based on spatiotemporal proximity and link stability to ensure context-relevant learning; and (iii) Context-Aware Synchronization, which adjusts aggregation frequency based on vehicular density and mobility to improve energy efficiency and learning consistency. Extensive experiments demonstrate that PA-FL achieves an average accuracy of 87.08% ± 0.49, surpassing state-of-the-art FL baselines by over 13% in accuracy and 11% in F1 score. It reduces task failure rates across all proximity ranges and lowers per-round energy consumption to 0.038 J, achieving a 6× improvement in communication efficiency. Delay per communication round is also reduced to 0.85 seconds, supporting real-time responsiveness. These results validate PA-FL as a resilient and scalable framework for symbiotic FL where vehicles collaboratively learn from local context while contributing to global intelligence in AI-integrated, 6G-enabled vehicular edge environments.

Abstract In the current Artificial Intelligence (AI) landscape, addressing explainability and interpretability in Machine Learning (ML) is of critical importance. In fact, the vast majority of works on AI focus on Deep Neural Networks (DNNs), which are not interpretable, as they are extremely hard to inspect and understand for humans. This is a crucial disadvantage of these methods, which hinders their trustability in high-stakes scenarios. On the other hand, interpretable models are considerably easier to inspect, which allows humans to test them exhaustively, and thus trust them. While the fields of eXplainable Artificial Intelligence (XAI) and Interpretable Artificial Intelligence (IAI) are progressing in supervised settings, the field of Interpretable Reinforcement Learning (IRL) is falling behind. Several approaches leveraging Decision Trees (DTs) for IRL have been proposed in recent years. However, all of them use goal-directed optimization methods, which may have limited exploration capabilities. In this work, we extend a previous study on the applicability of Quality–Diversity (QD) algorithms to the optimization of DTs for IRL. We test the methods on two well-known Reinforcement Learning (RL) benchmark tasks from OpenAI Gym, comparing their results in terms of score and “illumination” patterns. We show that using QD algorithms is an effective way to explore the search space of IRL models. Moreover, we find that, in the context of DTs for IRL, QD approaches based on MAP-Elites (ME) and its variant Covariance Matrix Adaptation MAP-Elites (CMA-ME) can significantly improve convergence speed over the goal-directed approaches.

Artificial intelligence methods are being increasingly applied across various domains, but their often opaque nature has raised concerns about accountability and trust. In response, the field of explainable AI (XAI) has emerged to address the need for human-understandable AI systems. Evolutionary computation (EC), a family of powerful optimization and learning algorithms, offers significant potential to contribute to XAI, and vice versa. This article provides an introduction to XAI and reviews current techniques for explaining machine learning (ML) models. We then explore how EC can be leveraged in XAI and examine existing XAI approaches that incorporate EC techniques. Furthermore, we discuss the application of XAI principles within EC itself, investigating how these principles can illuminate the behavior and outcomes of EC algorithms, their (automatic) configuration, and the underlying problem landscapes they optimize. Finally, we discuss open challenges in XAI and highlight opportunities for future research at the intersection of XAI and EC. Our goal is to demonstrate EC’s suitability for addressing current explainability challenges and to encourage further exploration of these methods, ultimately contributing to the development of more understandable and trustworthy ML models and EC algorithms.

No abstract available.

While excellent in transfer learning, Vision-Language models (VLMs) come with high computational costs due to their large number of parameters. To address this issue, removing parameters via model pruning is a viable solution. However, existing techniques for VLMs are task-specific, and thus require pruning the network from scratch for each new task of interest. In this work, we explore a new direction: Task-Agnostic Vision-Language Pruning (TA-VLP). Given a pretrained VLM, the goal is to find a unique pruned counterpart transferable to multiple unknown downstream tasks. In this challenging setting, the transferable representations already encoded in the pretrained model are a key aspect to preserve. Thus, we propose Multimodal Flow Pruning (MULTIFLOW), a first, gradient-free, pruning framework for TA-VLP where: (i) the importance of a parameter is expressed in terms of its magnitude and its information flow, by incorporating the saliency of the neurons it connects; and (ii) pruning is driven by the emergent (multimodal) distribution of the VLM parameters after pretraining. We benchmark eight state-of-the-art pruning algorithms in the context of TA-VLP, experimenting with two VLMs, three vision-language tasks, and three pruning ratios. Our experimental results show that MULTIFLOW outperforms recent sophisticated, combinatorial competitors in the vast majority of the cases, paving the way towards addressing TA- VLP. The code is publicly available at https://github.com/FarinaMatteo/multiflow.

In the past few years, Federated Learning (FL) has emerged as an effective approach for training neural networks (NNs) over a computing network while preserving data privacy. Most of the existing FL approaches require the user to define a priori the same structure for all the NNs running on the clients, along with an explicit aggregation procedure. This can be a limiting factor in cases where pre-defining such algorithmic details is difficult. To overcome these issues, we propose a novel approach to FL, which leverages Neuroevolution running on the clients. This implies that the NN structures may be different across clients, hence providing better adaptation to the local data. Furthermore, in our approach, the aggregation is implicitly accomplished on the client side by exploiting the information about the models used on the other clients, thus allowing the emergence of optimal NN architectures without needing an explicit aggregation. We test our approach on three datasets, showing that very compact NNs can be obtained without significant drops in performance compared to canonical FL. Moreover, we show that such compact structures allow for a step towards explainability, which is highly desirable in domains such as digital health, from which the tested datasets come.

One of the most striking capabilities behind the learning mechanisms of the brain is the adaptation, through structural and functional plasticity, of its synapses. While synapses have the fundamental role of transmitting information across the brain, several studies show that it is the neuron activations that produce changes on synapses. Yet, most plasticity models devised for artificial Neural Networks (NNs), e.g., the ABCD rule, focus on synapses, rather than neurons, therefore optimizing synaptic-specific Hebbian parameters. This approach, however, increases the complexity of the optimization process since each synapse is associated to multiple Hebbian parameters. To overcome this limitation, we propose a novel plasticity model, called Neuron-centric Hebbian Learning (NcHL), where optimization focuses on neuron- rather than synaptic-specific Hebbian parameters. Compared to the ABCD rule, NcHL reduces the parameters from 5W to 5N, being W and N the number of weights and neurons, and usually N « W. We also devise a "weightless" NcHL model, which requires less memory by approximating the weights based on a record of neuron activations. Our experiments on two robotic locomotion tasks reveal that NcHL performs comparably to the ABCD rule, despite using up to ~ 97 times less parameters, thus allowing for scalable plasticity.

Reinforcement learning (RL) algorithms often face challenges in efficiently learning effective policies for sparse-reward multi-goal robot manipulation tasks, thus requiring a vast amount of experiences. The state-of-the-art algorithm in the field, Hindsight Experience Replay (HER), addresses this issue by using failed trajectories and replacing the desired goal with hindsight goals. However, HER performs poorly when the desired goal is distant from the initial state. To address this limitation, Hindsight Goal Generation (HGG) has been proposed, which generates a curriculum of goals from already visited states. This curriculum generation is based on a single objective, and does not take obstacles into account. Here, we make a step forward by proposing Multi-Objective Evolutionary Hindsight Experience Replay (MOEHER), a novel curriculum RL algorithm that reformulates curriculum generation considering multiple objectives and obstacles. MOEHER utilizes NSGA-II to generate a curriculum that is optimized w.r.t. four objectives, namely the Q-function, the goal-proximity function, and two distance metrics, while simultaneously satisfying constraints on the obstacles. We evaluate MOEHER on four different sparse-reward robot manipulation tasks, with and without obstacles, and compare it with HER and HGG. The results demonstrate that MOEHER surpasses or performs on par with these methods on the tested tasks.

Vision-Language Models seamlessly discriminate among arbitrary semantic categories, yet they still suffer from poor generalization when presented with challenging examples. For this reason, Episodic Test-Time Adaptation (TTA) strategies have recently emerged as powerful techniques to adapt VLMs in the presence of a single unlabeled image. The recent literature on TTA is dominated by the paradigm of prompt tuning by Marginal Entropy Minimization, which, relying on online backpropagation, inevitably slows down inference while increasing memory. In this work, we theoretically investigate the properties of this approach and unveil that a surprisingly strong TTA method lies dormant and hidden within it. We term this approach ZERO (TTA with “zero” temperature), whose design is both incredibly effective and frustratingly simple: augment N times, predict, retain the most confident predictions, and marginalize after setting the Softmax temperature to zero. Remarkably, ZERO requires a single batched forward pass through the vision encoder only and no backward passes. We thoroughly evaluate our approach following the experimental protocol established in the literature and show that ZERO largely surpasses or compares favorably w.r.t. the state-of-the-art while being almost 10× faster and 13× more memory friendly than standard Test-Time Prompt Tuning. Thanks to its simplicity and comparatively negligible computation, ZERO can serve as a strong baseline for future work in this field. Code will be available.

Curriculum Reinforcement Learning (CRL) is an approach to facilitate the learning process of agents by structuring tasks in a sequence of increasing complexity. Despite its potential, many existing CRL methods struggle to efficiently guide agents toward desired outcomes, particularly in the absence of domain knowledge. This paper introduces DiCuRL (Diffusion Curriculum Reinforcement Learning), a novel method that leverages conditional diffusion models to generate curriculum goals. To estimate how close an agent is to achieving its goal, our method uniquely incorporates a $Q$-function and a trainable reward function based on Adversarial Intrinsic Motivation within the diffusion model. Furthermore, it promotes exploration through the inherent noising and denoising mechanism present in the diffusion models and is environment-agnostic. This combination allows for the generation of challenging yet achievable goals, enabling agents to learn effectively without relying on domain knowledge. We demonstrate the effectiveness of DiCuRL in three different maze environments and two robotic manipulation tasks simulated in MuJoCo, where it outperforms or matches nine state-of-the-art CRL algorithms from the literature.

Batteryless edge devices are extremely resource-constrained compared to traditional mobile platforms. Existing tiny deep neural network (DNN) inference solutions are problematic due to their slow and resource-intensive nature, rendering them unsuitable for batteryless edge devices. To address this problem, we propose a new approach to embedded intelligence, called Fast-Inf, which achieves extremely lightweight computation and minimal latency. Fast-Inf uses binary tree-based neural networks that are ultra-fast and energy-efficient due to their logarithmic time complexity. Additionally, Fast-Inf models can skip the leaf nodes when necessary, further minimizing latency without requiring any modifications to the model or retraining. Moreover, Fast-Inf models have significantly lower backup and runtime memory overhead. Our experiments on an MSP430FR5994 platform showed that Fast-Inf can achieve ultra-fast and energy-efficient inference (up to 700x speedup and reduced energy) compared to a conventional DNN.

Abstract Several constrained optimization problems have been adequately solved over the years thanks to the advances in the area of metaheuristics. Nevertheless, the question as to which search logic performs better on constrained optimization often arises. In this paper, we present Dual Search Optimization (DSO), a co-evolutionary algorithm that includes an adaptive penalty function to handle constrained problems. Compared to other self-adaptive metaheuristics, one of the main advantages of DSO is that it is able auto-construct its own perturbation logics, i.e., the ways solutions are modified to create new ones during the optimization process. This is accomplished by co-evolving the solutions (encoded as vectors of integer/real values) and perturbation strategies (encoded as Genetic Programming trees), in order to adapt the search to the problem. In addition to that, the adaptive penalty function allows the algorithm to handle constraints very effectively, yet with a minor additional algorithmic overhead. We compare DSO with several algorithms from the state-of-the-art on two sets of problems, namely: (1) seven well-known constrained engineering design problems and (2) the CEC 2017 benchmark for constrained optimization. Our results show that DSO can achieve state-of-the-art performances, being capable to automatically adjust its behavior to the problem at hand.

Explainable Artificial Intelligence (XAI) has recently emerged as one of the most active areas of research in AI. While Evolutionary Computation (EC) is also a very active research area, the intersection between XAI and EC is still rather unexplored. This topic was the subject of our Workshops on Evolutionary Computing and Explainable Artificial Intelligence(ECXAI), organized at GECCO 2022 and GECCO 2023. This special issue collects four articles further exploring the intersection between XAI and EC, including both the use of EC for XAI as well as the use of explainability techniques to better understand EC methods.

Ultrasound localization microscopy (ULM) allows for the generation of super-resolved (SR) images of the vasculature by precisely localizing intravenously injected microbubbles. Although SR images may be useful for diagnosing and treating patients, their use in the clinical context is limited by the need for prolonged acquisition times and high frame rates. The primary goal of our study is to relax the requirement of high frame rates to obtain SR images. To this end, we propose a new time-efficient ULM (TEULM) pipeline built on a cutting-edge interpolation method. More specifically, we suggest employing Radial Basis Functions (RBFs) as interpolators to estimate the missing values in the 2-dimensional (2D) spatio-temporal structures. To evaluate this strategy, we first mimic the data acquisition at a reduced frame rate by applying a down-sampling (DS = 2, 4, 8, and 10) factor to high frame rate ULM data. Then, we up-sample the data to the original frame rate using the suggested interpolation to reconstruct the missing frames. Finally, using both the original high frame rate data and the interpolated one, we reconstruct SR images using the ULM framework steps. We evaluate the proposed TEULM using four in vivo datasets, a Rat brain (dataset A), a Rat kidney (dataset B), a Rat tumor (dataset C) and a Rat brain bolus (dataset D), interpolating at the in-phase and quadrature (IQ) level. Results demonstrate the effectiveness of TEULM in recovering vascular structures, even at a DS rate of 10 (corresponding to a frame rate of sub-100Hz). In conclusion, the proposed technique is successful in reconstructing accurate SR images while requiring frame rates of one order of magnitude lower than standard ULM.

With the advent of cheap, miniaturized electronics, ubiquitous networking has reached an unprecedented level of complexity, scale and heterogeneity, becoming the core of several modern applications such as smart industry, smart buildings and smart cities. A crucial element for network performance is the protocol stack, namely the sets of rules and data formats that determine how the nodes in the network exchange information. A great effort has been put to devise formal techniques to synthesize (offline) network protocols, starting from system specifications and strict assumptions on the network environment. However, offline design can be hard to apply in the most modern network applications, either due to numerical complexity, or to the fact that the environment might be unknown and the specifications might not available. In these cases, online protocol design and adaptation has the potential to offer a much more scalable and robust solution. Nevertheless, so far only a few attempts have been done towards online automatic protocol design. These approaches, however, typically require a central coordinator, or need to build and update a model of the environment, which adds complexity. Here, instead, we envision a protocol as an emergent property of a network, obtained by an environment-driven Distributed Hill Climbing (DHC) algorithm that uses node-local reinforcement signals to evolve, at runtime and without any central coordination, a network protocol from scratch, without needing a model of the environment. We test this approach with a 3-state Time Division Multiple Access (TDMA) Medium Access Control (MAC) protocol and we observe its emergence in networks of various scales and with various settings. We also show how DHC can reach different trade-offs in terms of energy consumption and protocol performance.

In the last three decades, the field of computational intelligence has seen a profusion of population‐based metaheuristics applied to a variety of problems, where they achieved state‐of‐the‐art results. This remarkable growth has been fuelled and, to some extent, exacerbated by various sources of inspiration and working philosophies, which have been thoroughly reviewed in several recent survey papers. However, the present survey addresses an important gap in the literature. Here, we reflect on a systematic categorisation of what we call “lightweight” metaheuristics, i.e., optimisation algorithms characterised by purposely limited memory and computational requirements. We focus mainly on two classes of lightweight algorithms: single‐solution metaheuristics and “compact” optimisation algorithms. Our analysis is mostly focused on single‐objective continuous optimisation. We provide an updated and unified view of the most important achievements in the field of lightweight metaheuristics, background concepts, and most important applications. We then discuss the implications of these algorithms and the main open questions and suggest future research directions.

Finding the most influential nodes in a network is a computationally hard problem with several possible applications in various kinds of network-based problems. While several methods have been proposed for tackling the influence maximisation (IM) problem, their runtime typically scales poorly when the network size increases. Here, we propose an original method, based on network downscaling, that allows a multi-objective evolutionary algorithm (MOEA) to solve the IM problem on a reduced scale network, while preserving the relevant properties of the original network. The downscaled solution is then upscaled to the original network, using a mechanism based on centrality metrics such as PageRank. Our results on eight large networks (including two with $\sim$50k nodes) demonstrate the effectiveness of the proposed method with a more than 10-fold runtime gain compared to the time needed on the original network, and an up to $82\%$ time reduction compared to CELF.

The evolution of symbolic communication is a longstanding open research question in biology. While some theories suggest that it originated from sub-symbolic communication (i.e., iconic or indexical), little experimental evidence exists on how organisms can actually evolve to define a shared set of symbols with unique interpretable meaning, thus being capable of encoding and decoding discrete information. Here, we use a simple synthetic model composed of sender and receiver agents controlled by Continuous-Time Recurrent Neural Networks, which are optimized by means of neuro-evolution. We characterize signal decoding as either regression or classification, with limited and unlimited signal amplitude. First, we show how this choice affects the complexity of the evolutionary search, and leads to different levels of generalization. We then assess the effect of noise, and test the evolved signaling system in a referential game. In various settings, we observe agents evolving to share a dictionary of symbols, with each symbol spontaneously associated to a 1-D unique signal. Finally, we analyze the constellation of signals associated to the evolved signaling systems and note that in most cases these resemble a Pulse Amplitude Modulation system.

Designing optimal soft modular robots is difficult, due to non-trivial interactions between morphology and controller. Evolutionary algorithms (EAs), combined with physical simulators, represent a valid tool to overcome this issue. In this work, we investigate algorithmic solutions to improve the Quality Diversity of co-evolved designs of Tensegrity Soft Modular Robots (TSMRs) for two robotic tasks, namely goal reaching and squeezing trough a narrow passage. To this aim, we use three different EAs, i.e., MAP-Elites and two custom algorithms: one based on Viability Evolution (ViE) and NEAT (ViE-NEAT), the other named Double Map MAP-Elites (DM-ME) and devised to seek diversity while co-evolving robot morphologies and neural network (NN)-based controllers. In detail, DM-ME extends MAP-Elites in that it uses two distinct feature maps, referring to morphologies and controllers respectively, and integrates a mechanism to automatically define the NN-related feature descriptor. Considering the fitness, in the goal-reaching task ViE-NEAT outperforms MAP-Elites and results equivalent to DM-ME. Instead, when considering diversity in terms of "illumination" of the feature space, DM-ME outperforms the other two algorithms on both tasks, providing a richer pool of possible robotic designs, whereas ViE-NEAT shows comparable performance to MAP-Elites on goal reaching, although it does not exploit any map.

A fundamental aspect of learning in biological neural networks is the plasticity property which allows them to modify their configurations during their lifetime. Hebbian learning is a biologically plausible mechanism for modeling the plasticity property in artificial neural networks (ANNs), based on the local interactions of neurons. However, the emergence of a coherent global learning behavior from local Hebbian plasticity rules is not very well understood. The goal of this work is to discover interpretable local Hebbian learning rules that can provide autonomous global learning. To achieve this, we use a discrete representation to encode the learning rules in a finite search space. These rules are then used to perform synaptic changes, based on the local interactions of the neurons. We employ genetic algorithms to optimize these rules to allow learning on two separate tasks (a foraging and a prey-predator scenario) in online lifetime learning settings. The resulting evolved rules converged into a set of well-defined interpretable types, that are thoroughly discussed. Notably, the performance of these rules, while adapting the ANNs during the learning tasks, is comparable to that of offline learning methods such as hill climbing.

Routing plays a fundamental role in network applications, but it is especially challenging in Delay Tolerant Networks (DTNs). These are a kind of mobile ad hoc networks made of, e.g., (possibly, unmanned) vehicles and humans where, despite a lack of continuous connectivity, data must be transmitted while the network conditions change due to the nodes’ mobility. In these contexts, routing is NP-hard and is usually solved by heuristic “store and forward” replication-based approaches, where multiple copies of the same message are moved and stored across nodes in the hope that at least one will reach its destination. Still, the existing routing protocols produce relatively low delivery probabilities. Here, we genetically improve two routing protocols widely adopted in DTNs, namely, Epidemic and PRoPHET, in the attempt to optimize their delivery probability. First, we dissect them into their fundamental components, i.e., functionalities such as checking if a node can transfer data, or sending messages to all connections. Then, we apply Genetic Improvement (GI) to manipulate these components as terminal nodes of evolving trees. We apply this methodology, in silico, to six test cases of urban networks made of hundreds of nodes and find that GI produces consistent gains in delivery probability in four cases. We then verify if this improvement entails a worsening of other relevant network metrics, such as latency and buffer time. Finally, we compare the logics of the best evolved protocols with those of the baseline protocols, and we discuss the generalizability of the results across test cases.

As Autonomous vehicles (AVs) are entering shared roads, the challenge of designing and implementing a completely autonomous vehicle is still open. Aside from technological issues regarding how to manage the complexity of the environment, AVs raise difficult legal issues and ethical dilemmas, especially in unavoidable accident scenarios. In this context, a vast speculation depicting moral dilemmas has developed in recent years. A new perspective was proposed: an “Ethical Knob” (EK), enabling passengers to ethically customise their AVs, namely, to choose between different settings corresponding to different moral approaches or principles. In this contribution we explore how an AV can automatically learn to determine the value of its “Ethical Knob” in order to achieve a trade-off between the ethical preferences of passengers and social values, learning from experienced instances of collision. To this end, we propose a novel approach based on a genetic algorithm to optimize a population of neural networks. We report a detailed description of simulation experiments as well as possible applications.

The plasticity property of biological neural networks allows them to perform learning and optimize their behavior by changing their configuration. Inspired by biology, plasticity can be modeled in artificial neural networks by using Hebbian learning rules, i.e. rules that update synapses based on the neuron activations and reinforcement signals. However, the distal reward problem arises when the reinforcement signals are not available immediately after each network output to associate the neuron activations that contributed to receiving the reinforcement signal. In this work, we extend Hebbian plasticity rules to allow learning in distal reward cases. We propose the use of neuron activation traces (NATs) to provide additional data storage in each synapse to keep track of the activation of the neurons. Delayed reinforcement signals are provided after each episode relative to the networks' performance during the previous episode. We employ genetic algorithms to evolve delayed synaptic plasticity (DSP) rules and perform synaptic updates based on NATs and delayed reinforcement signals. We compare DSP with an analogous hill climbing algorithm that does not incorporate domain knowledge introduced with the NATs, and show that the synaptic updates performed by the DSP rules demonstrate more effective training performance relative to the HC algorithm.

Many real-world control and classification tasks involve a large number of features. When artificial neural networks (ANNs) are used for modeling these tasks, the network architectures tend to be large. Neuroevolution is an effective approach for optimizing ANNs; however, there are two bottlenecks that make their application challenging in case of high-dimensional networks using direct encoding. First, classic evolutionary algorithms tend not to scale well for searching large parameter spaces; second, the network evaluation over a large number of training instances is in general time-consuming. In this work, we propose an approach called the Limited Evaluation Cooperative Co-evolutionary Differential Evolution algorithm (LECCDE) to optimize high-dimensional ANNs.

Evolutionary algorithms are increasingly being applied to problems that are too computationally expensive to run on a single personal computer due to costly fitness function evaluations and/or large numbers of fitness evaluations. Here, we introduce the Seamless Peer And Cloud Evolution (SPACE) framework, which leverages bleeding edge web technologies to allow the computational resources necessary for running large scale evolutionary experiments to be made available to amateur and professional researchers alike, in a scalable and cost-effective manner, directly from their web browsers. The SPACE framework accomplishes this by distributing fitness evaluations across a heterogeneous pool of cloud compute nodes and peer computers. As a proof of concept, this framework has been attached to the \hbox{RoboGen\texttrademark} open-source platform for the co-evolution of robot bodies and brains, but importantly the framework has been built in a modular fashion such that it can be easily coupled with other evolutionary computation systems.

The performance of evolutionary algorithms can be heavily undermined when constraints limit the feasible areas of the search space. For instance, while covariance matrix adaptation evolution strategy (CMA-ES) is one of the most efficient algorithms for unconstrained optimization problems, it cannot be readily applied to constrained ones. Here, we used concepts from memetic computing, i.e., the harmonious combination of multiple units of algorithmic information, and viability evolution, an alternative abstraction of artificial evolution, to devise a novel approach for solving optimization problems with inequality constraints. Viability evolution emphasizes the elimination of solutions that do not satisfy viability criteria, which are defined as boundaries on objectives and constraints. These boundaries are adapted during the search to drive a population of local search units, based on CMA-ES, toward feasible regions. These units can be recombined by means of differential evolution operators. Of crucial importance for the performance of our method, an adaptive scheduler toggles between exploitation and exploration by selecting to advance one of the local search units and/or recombine them. The proposed algorithm can outperform several state-of-the-art methods on a diverse set of benchmark and engineering problems, both for quality of solutions and computational resources needed.

Wireless sensor network (WSN) routing protocols, e.g., the Collection Tree Protocol (CTP), are designed to adapt in an ad-hoc fashion to the quality of the environment. WSNs thus have high internal dynamics and complex global behavior. Classical techniques for performance evaluation (such as testing or verification) fail to uncover the cases of extreme behavior which are most interesting to designers. We contribute a practical framework for performance evaluation of WSN protocols. The framework is based on multi-objective optimization, coupled with protocol simulation and evaluation of performance factors. For evaluation, we consider the two crucial functional and non-functional performance factors of a WSN, respectively: the ratio of data delivery from the network (DDR), and the total energy expenditure of the network (COST). We are able to discover network topological configurations over which CTP has unexpectedly low DDR and/or high COST performance, and expose full Pareto fronts which show what the possible performance tradeoffs for CTP are in terms of these two performance factors. Eventually, Pareto fronts allow us to bound the state space of the WSN, a fact which provides essential knowledge to WSN protocol designers.

We propose Multi-Strategy Coevolving Aging Particles (MS-CAP), a novel population-based algorithm for black-box optimization. In a memetic fashion, MS-CAP combines two components with complementary algorithm logics. In the first stage, each particle is perturbed independently along each dimension with a progressively shrinking (decaying) radius, and attracted towards the current best solution with an increasing force. In the second phase, the particles are mutated and recombined according to a multi-strategy approach in the fashion of the ensemble of mutation strategies in Differential Evolution. The proposed algorithm is tested, at different dimensionalities, on two complete black-box optimization benchmarks proposed at the Congress on Evolutionary Computation 2010 and 2013. To demonstrate the applicability of the approach, we also test MS-CAP to train a Feedforward Neural Network modeling the kinematics of an 8-link robot manipulator. The numerical results show that MS-CAP, for the setting considered in this study, tends to outperform the state-of-the-art optimization algorithms on a large set of problems, thus resulting in a robust and versatile optimizer.

The super-fit scheme, consisting of injecting an individual with high fitness into the initial population of an algorithm, has shown to be a simple and effective way to enhance the algorithmic performance of the population-based algorithm. Whether the super-fit individual is based on some prior knowledge on the optimization problem or is derived from an initial step of pre-processing, e.g. a local search, this mechanism has been applied successfully in various examples of evolutionary and swarm intelligence algorithms. This paper presents an unconventional application of this super-fit scheme, where the super-fit individual is obtained by means of the Covariance Adaptation Matrix Evolution Strategy (CMA-ES), and fed to a single solution local search which perturbs iteratively each variable. Thus, compared to other super-fit schemes, the roles of super-fit individual generator and global optimizer are switched. To prevent premature convergence, the local search employs a re-sampling mechanism which inherits parts of the best individual while randomly sampling the remaining variables. We refer to such local search as Re-sampled Inheritance Search (RIS). Tested on the CEC 2013 optimization benchmark, the proposed algorithm, named CMA-ES-RIS, displays a respectable performance and a good balance between exploration and exploitation, resulting into a versatile and robust optimization tool.

This paper proposes an algorithm to solve the CEC2013 benchmark. The algorithm, namely Super-fit Multicriteria Adaptive Differential Evolution (SMADE), is a Memetic Computing approach based on the hybridization of two algorithmic schemes according to a super-fit memetic logic. More specifically, the Covariance Matrix Adaptive Evolution Strategy (CMAES), run at the beginning of the optimization process, is used to generate a solution with a high quality. This solution is then injected into the population of a modified Differential Evolution, namely Multicriteria Adaptive Differential Evolution (MADE). The improved solution is super-fit as it supposedly exhibits a performance a way higher than the other population individuals. The super-fit individual then leads the search of the MADE scheme towards the optimum. Unimodal or mildly multimodal problems, even when non-separable and ill-conditioned, tend to be solved during the early stages of the optimization by the CMAES. Highly multi-modal optimization problems are efficiently tackled by SMADE since the MADE algorithm (as well as other Differential Evolution schemes) appears to work very well when the search is led by a super-fit individual.

This paper introduces two lightweight variants of ISPO, a Single Particle Optimization algorithm recently proposed in the literature. The goal of this work is to improve upon the performance of the original ISPO, still bearing in mind its admirable algorithmic simplicity. The first variant, namely ISPOrestart, combines in a memetic fashion the logics of ISPO with a partial restart mechanism similar to the binomial crossover typically used in Differential Evolution. The second variant, named VISPO, builds on top of the restart process a very simple learning stage which tries to adapt the algorithm behaviour to the (non)-separability of the problem. Numerical results obtained on three complete optimization benchmarks show that not only the two algorithms are able to improve, incrementally, upon the performance of ISPO, but also they show respectable performance in comparison with modern complex state-of-the-art methods, especially when the problem dimensionality increases.

The maturing field of Wireless Sensor Networks (WSN) results in long-lived deployments that produce large amounts of sensor data. Lightweight online on-mote processing may improve the usage of their limited resources, such as energy, by transmitting only unexpected sensor data (anomalies). We detect anomalies by analyzing sensor reading predictions from a linear model. We use Recursive Least Squares (RLS) to estimate the model parameters, because for large datasets the standard Linear Least Squares Estimation (LLSE) is not resource friendly. We evaluate the use of fixed-point RLS with adaptive thresholding, and its application to anomaly detection in embedded systems. We present an extensive experimental campaign on generated and real-world datasets, with floating-point RLS, LLSE, and a rule-based method as benchmarks. The methods are evaluated on prediction accuracy of the models, and on detection of anomalies, which are injected in the generated dataset. The experimental results show that the proposed algorithm is comparable, in terms of prediction accuracy and detection performance, to the other LS methods. However, fixed-point RLS is efficiently implement able in embedded devices. The presented method enables online on-mote anomaly detection with results comparable to offline LS methods.

Differential Evolution is a population based stochastic algorithm with less number of parameters to tune. However, the performance of DE is sensitive to the mutation and crossover strategies and their associated parameters. To obtain optimal performance, DE requires time consuming trial and error parameter tuning. To overcome the computationally expensive parameter tuning different adaptive/self-adaptive techniques have been proposed. Recently the idea of ensemble strategies in DE has been proposed and favorably compared with some of the state-of-the-art self-adaptive techniques. Compact Differential Evolution (cDE) is modified version of DE algorithm which can be effectively used to solve real world problems where sufficient computational resources are not available. cDE can be implemented on devices such as micro controllers or Graphics Processing Units (GPUs) which have limited memory. In this paper we introduced the idea of ensemble into cDE to improve its performance. The proposed algorithm is tested on the 30D version of 14 benchmark problems of Conference on Evolutionary Computation (CEC) 2005. The employment of ensemble strategies for the cDE algorithms appears to be beneficial and leads, for some problems, to competitive results with respect to the-state-of the-art DE based algorithms.