Surgical robot task automation has recently attracted great attention due to its potential to benefit both surgeons and patients. Reinforcement learning (RL) based approaches have demonstrated promising ability to provide solutions to automated surgical manipulations on various tasks. To address the exploration challenge, expert demonstrations can be utilized to enhance the learning efficiency via imitation learning (IL) approaches. However, the successes of such methods normally rely on both states and action labels. Unfortunately action labels can be hard to capture or their manual annotation is prohibitively expensive owing to the requirement for expert knowledge. It therefore remains an appealing and open problem to leverage expert demonstrations composed of pure states in RL. In this work, we present an actor-critic RL framework, termed AC-SSIL, to overcome this challenge of learning with state-only demonstrations collected by following an unknown expert policy. It adopts a self-supervised IL method, dubbed SSIL, to effectively incorporate demonstrated states into RL paradigms by retrieving from demonstrates the nearest neighbours of the query state and utilizing the bootstrapping of actor networks. We showcase through experiments on an open-source surgical simulation platform that our method delivers remarkable improvements over the RL baseline and exhibits comparable performance against action based IL methods, which implies the efficacy and potential of our method for expert demonstration-guided learning scenarios.

We present a new approach for solving systems of linear equations with tensor networks based on the quantum HHL algorithm. We first develop a novel HHL in the qudits formalism, the generalization of qubits, and then transform its operations into an equivalent classical HHL, taking advantage of the non-unitary operations that they can apply. The main novelty of this proposal is to perform a classical simulation as efficiently as possible of the HHL to benchmark the algorithm steps according to its input parameters and the input matrix. We apply this algorithm to three simulation problems, comparing it with an exact inversion algorithm, and we compare its performance against an implementation of the original HHL simulated in the Qiskit framework, providing both codes. Our results show that our approach can achieve a promising performance in computational efficiency to simulate HHL process without quantum noise, providing a lower bound.

In this article, a benchmark for real-world bin packing problems is proposed. This dataset consists of 12 instances of varying levels of complexity regarding size (with the number of packages ranging from 38 to 53) and user-defined requirements. In fact, several real-world-oriented restrictions were taken into account to build these instances: i) item and bin dimensions, ii) weight restrictions, iii) affinities among package categories iv) preferences for package ordering and v) load balancing. Besides the data, we also offer an own developed Python script for the dataset generation, coined Q4RealBPP-DataGen. The benchmark was initially proposed to evaluate the performance of quantum solvers. Therefore, the characteristics of this set of instances were designed according to the current limitations of quantum devices. Additionally, the dataset generator is included to allow the construction of general-purpose benchmarks. The data introduced in this article provides a baseline that will encourage quantum computing researchers to work on real-world bin packing problems.

Herbicide field trials require accurate identification of plant species and assessment of herbicide-induced damage across diverse environments. While general-purpose vision foundation models have shown promising results in complex visual domains, their performance can be limited in agriculture, where fine-grained distinctions between species and damage types are critical. In this work, we adapt a general-purpose vision foundation model to herbicide trial characterization. Trained using a self-supervised learning approach on a large, curated agricultural dataset, the model learns rich and transferable representations optimized for herbicide trials images. Our domain-specific model significantly outperforms the best general-purpose foundation model in both species identification (F1 score improvement from 0.91 to 0.94) and damage classification (from 0.26 to 0.33). Under unseen conditions (new locations and other time), it achieves even greater gains (species identification from 0.56 to 0.66; damage classification from 0.17 to 0.27). In domain-shift scenarios, such as drone imagery, it maintains strong performance (species classification from 0.49 to 0.60). Additionally, we show that domain-specific pretraining enhances segmentation accuracy, particularly in low-annotation regimes. An annotation-efficiency analysis reveals that, under unseen conditions, the domain-specific model achieves 5.4% higher F1 score than the general-purpose model, while using 80% fewer labeled samples. These results demonstrate the generalization capabilities of domain-specific foundation models and their potential to significantly reduce manual annotation efforts, offering a scalable and automated solution for herbicide trial analysis.

Large Language Models (LLM) have achieved remarkable performance across a large number of tasks, but face critical deployment and usage barriers due to substantial computational requirements. Model compression methods, which aim to reduce model size while preserving its capacity, are an important means to mitigate these issues. Promising approaches along these lines, such as structured pruning, typically require costly empirical search for optimal variants and may run the risk of ignoring better solutions. In this work we introduce GeLaCo, an evolutionary approach to LLM compression via layer collapse. Our approach supports an efficient exploration of the compression solution space via population-based search and a module-wise similarity fitness function capturing attention, feed-forward, and hidden state representations. GeLaCo also supports both single and multi-objective evolutionary compression search, establishing the first Pareto frontier along compression and quality axes. We evaluate GeLaCo solutions via both perplexity-based and generative evaluations over foundational and instruction-tuned models, outperforming state-of-the-art alternatives.

In recent years, Artificial Intelligence (AI) models have achieved remarkable success across various domains, yet challenges persist in two critical areas: ensuring robustness against uncertain inputs and drastically increasing model efficiency during training and inference. Spiking Neural Networks (SNNs), inspired by biological systems, offer a promising avenue for overcoming these limitations. By operating in an event-driven manner, SNNs achieve low energy consumption and can naturally implement biological methods known for their high noise tolerance. In this work, we explore the potential of the spiking Forward-Forward Algorithm (FFA) to address these challenges, leveraging its representational properties for both Out-of-Distribution (OoD) detection and interpretability. To achieve this, we exploit the sparse and highly specialized neural latent space of FF networks to estimate the likelihood of a sample belonging to the training distribution. Additionally, we propose a novel, gradient-free attribution method to detect features that drive a sample away from class distributions, addressing the challenges posed by the lack of gradients in most visual interpretability methods for spiking models. We evaluate our OoD detection algorithm on well-known image datasets (e.g., Omniglot, Not-MNIST, CIFAR10), outperforming previous methods proposed in the recent literature for OoD detection in spiking networks. Furthermore, our attribution method precisely identifies salient OoD features, such as artifacts or missing regions, hence providing a visual explanatory interface for the user to understand why unknown inputs are identified as such by the proposed method.

Quantum and classical machine learning have been naturally connected through kernel methods, which have also served as proof-of-concept for quantum advantage. Quantum embeddings encode classical data into quantum feature states, enabling the construction of embedding quantum kernels (EQKs) by measuring vector similarities and projected quantum kernels (PQKs) through projections of these states. However, in both approaches, the model is influenced by the choice of the embedding. In this work, we propose using the training of a quantum neural network (QNN) to construct neural quantum kernels: both neural EQKs and neural PQKs, which are problem-inspired kernels. Unlike previous methods in the literature, our approach requires the kernel matrix to be constructed only once. We present several strategies for constructing neural quantum kernels and propose a scalable method to train an $n$-qubit data re-uploading quantum neural network (QNN). We provide numerical evidence of the performance of these models under noisy conditions. Additionally, we demonstrate how neural quantum kernels can alleviate exponential concentration and enhance generalization capabilities compared to problem-agnostic kernels, positioning them as a scalable and robust solution for quantum machine learning applications.

Quantum thermalization in contemporary quantum devices, in particular quantum computers, has recently attracted significant theoretical interest. Unusual thermalization processes, such as the Quantum Mpemba Effect (QME), have been explored theoretically. However, there is a shortage of experimental results due to the difficulty in preparing thermal states. In this paper, we propose a protocol to indirectly address this challenge. Moreover, we experimentally validate our protocol using IBM quantum devices, providing results for unusual relaxation in equidistant quenches as predicted for the IBM qubit. We also assess the formalism introduced for the QME, obtaining results consistent with the theoretical predictions. This demonstration underscores that our protocol can provide an alternative way of studying thermal states physics when their direct preparation may be too difficult.

In recent years, bio-inspired optimization methods, which mimic biological processes to solve complex problems, have gained popularity in recent literature. The proliferation of proposals prove the growing interest in this field. The increase in nature- and bio-inspired algorithms, applications, and guidelines highlights growing interest in this field. However, the exponential rise in the number of bio-inspired algorithms poses a challenge to the future trajectory of this research domain. Along the five versions of this document, the number of approaches grows incessantly, and where having a new biological description takes precedence over real problem-solving. This document presents two comprehensive taxonomies. One based on principles of biological similarity, and the other one based on operational aspects associated with the iteration of population models that initially have a biological inspiration. Therefore, these taxonomies enable researchers to categorize existing algorithmic developments into well-defined classes, considering two criteria: the source of inspiration, and the behavior exhibited by each algorithm. Using these taxonomies, we classify 518 algorithms based on nature-inspired and bio-inspired principles. Each algorithm within these categories is thoroughly examined, allowing for a critical synthesis of design trends and similarities, and identifying the most analogous classical algorithm for each proposal. From our analysis, we conclude that a poor relationship is often found between the natural inspiration of an algorithm and its behavior. Furthermore, similarities in terms of behavior between different algorithms are greater than what is claimed in their public disclosure: specifically, we show that more than one-fourth of the reviewed solvers are versions of classical algorithms. The conclusions from the analysis of the algorithms lead to several learned lessons.

Efficient packing of items into bins is a common daily task. Known as Bin Packing Problem, it has been intensively studied in the field of artificial intelligence, thanks to the wide interest from industry and logistics. Since decades, many variants have been proposed, with the three-dimensional Bin Packing Problem as the closest one to real-world use cases. We introduce a hybrid quantum-classical framework for solving real-world three-dimensional Bin Packing Problems (Q4RealBPP), considering different realistic characteristics, such as: i) package and bin dimensions, ii) overweight restrictions, iii) affinities among item categories and iv) preferences for item ordering. Q4RealBPP permits the solving of real-world oriented instances of 3dBPP, contemplating restrictions well appreciated by industrial and logistics sectors.

The expectations arising from the latest achievements in the quantum computing field are causing that researchers coming from classical artificial intelligence to be fascinated by this new paradigm. In turn, quantum computing, on the road towards usability, needs classical procedures. Hybridization is, in these circumstances, an indispensable step but can also be seen as a promising new avenue to get the most from both computational worlds. Nonetheless, hybrid approaches have now and will have in the future many challenges to face, which, if ignored, will threaten the viability or attractiveness of quantum computing for real-world applications. To identify them and pose pertinent questions, a proper characterization of the hybrid quantum computing field, and especially hybrid solvers, is compulsory. With this motivation in mind, the main purpose of this work is to propose a preliminary taxonomy for classifying hybrid schemes, and bring to the fore some questions to stir up researchers minds about the real challenges regarding the application of quantum computing.

Histopathology image classification is crucial for the accurate identification and diagnosis of various diseases but requires large and diverse datasets. Obtaining such datasets, however, is often costly and time-consuming due to the need for expert annotations and ethical constraints. To address this, we examine the suitability of different generative models and image selection approaches to create realistic synthetic histopathology image patches conditioned on class labels. Our findings highlight the importance of selecting an appropriate generative model type and architecture to enhance performance. Our experiments over the PCam dataset show that diffusion models are effective for transfer learning, while GAN-generated samples are better suited for augmentation. Additionally, transformer-based generative models do not require image filtering, in contrast to those derived from Convolutional Neural Networks (CNNs), which benefit from realism score-based selection. Therefore, we show that synthetic images can effectively augment existing datasets, ultimately improving the performance of the downstream histopathology image classification task.

One of the key challenges of Reinforcement Learning (RL) is the ability of agents to generalise their learned policy to unseen settings. Moreover, training RL agents requires large numbers of interactions with the environment. Motivated by the recent success of Offline RL and Imitation Learning (IL), we conduct a study to investigate whether agents can leverage offline data in the form of trajectories to improve the sample-efficiency in procedurally generated environments. We consider two settings of using IL from offline data for RL: (1) pre-training a policy before online RL training and (2) concurrently training a policy with online RL and IL from offline data. We analyse the impact of the quality (optimality of trajectories) and diversity (number of trajectories and covered level) of available offline trajectories on the effectiveness of both approaches. Across four well-known sparse reward tasks in the MiniGrid environment, we find that using IL for pre-training and concurrently during online RL training both consistently improve the sample-efficiency while converging to optimal policies. Furthermore, we show that pre-training a policy from as few as two trajectories can make the difference between learning an optimal policy at the end of online training and not learning at all. Our findings motivate the widespread adoption of IL for pre-training and concurrent IL in procedurally generated environments whenever offline trajectories are available or can be generated.

This study addresses the minor-embedding problem, which involves mapping the variables of an Ising model onto a quantum annealing processor. The primary motivation stems from the observed performance disparity of quantum annealers when solving problems suited to the processor's architecture versus those with non-hardware-native topologies. Our research has two main objectives: i) to analyze the impact of embedding quality on the performance of D-Wave Systems quantum annealers, and ii) to evaluate the quality of the embeddings generated by Minorminer, the standard minor-embedding technique in the quantum annealing literature, provided by D-Wave. Regarding the first objective, our experiments reveal a clear correlation between the average chain length of embeddings and the relative errors of the solutions sampled. This underscores the critical influence of embedding quality on quantum annealing performance. For the second objective, we evaluate Minorminer's embedding capabilities, the quality and robustness of its embeddings, and its execution-time performance. We also compare its performance with Clique Embedding, another algorithm developed by D-Wave, which is deterministic and designed to embed fully connected Ising models into quantum annealing processors, serving as a worst-case scenario. The results demonstrate that there is significant room for improvement for Minorminer, suggesting that more effective embedding strategies could lead to meaningful gains in quantum annealing performance.

Multitasking optimization is an incipient research area which is lately gaining a notable research momentum. Unlike traditional optimization paradigm that focuses on solving a single task at a time, multitasking addresses how multiple optimization problems can be tackled simultaneously by performing a single search process. The main objective to achieve this goal efficiently is to exploit synergies between the problems (tasks) to be optimized, helping each other via knowledge transfer (thereby being referred to as Transfer Optimization). Furthermore, the equally recent concept of Evolutionary Multitasking (EM) refers to multitasking environments adopting concepts from Evolutionary Computation as their inspiration for the simultaneous solving of the problems under consideration. As such, EM approaches such as the Multifactorial Evolutionary Algorithm (MFEA) has shown a remarkable success when dealing with multiple discrete, continuous, single-, and/or multi-objective optimization problems. In this work we propose a novel algorithmic scheme for Multifactorial Optimization scenarios - the Multifactorial Cellular Genetic Algorithm (MFCGA) - that hinges on concepts from Cellular Automata to implement mechanisms for exchanging knowledge among problems. We conduct an extensive performance analysis of the proposed MFCGA and compare it to the canonical MFEA under the same algorithmic conditions and over 15 different multitasking setups (encompassing different reference instances of the discrete Traveling Salesman Problem). A further contribution of this analysis beyond performance benchmarking is a quantitative examination of the genetic transferability among the problem instances, eliciting an empirical demonstration of the synergies emerged between the different optimization tasks along the MFCGA search process.

Metric learning has become an attractive field for research on the latest years. Loss functions like contrastive loss, triplet loss or multi-class N-pair loss have made possible generating models capable of tackling complex scenarios with the presence of many classes and scarcity on the number of images per class not only work to build classifiers, but to many other applications where measuring similarity is the key. Deep Neural Networks trained via metric learning also offer the possibility to solve few-shot learning problems. Currently used state of the art loss functions such as triplet and contrastive loss functions, still suffer from slow convergence due to the selection of effective training samples that has been partially solved by the multi-class N-pair loss by simultaneously adding additional samples from the different classes. In this work, we extend triplet and multiclass-N-pair loss function by proposing the constellation loss metric where the distances among all class combinations are simultaneously learned. We have compared our constellation loss for visual class embedding showing that our loss function over-performs the other methods by obtaining more compact clusters while achieving better classification results.

Transfer Optimization has gained a remarkable attention from the Swarm and Evolutionary Computation community in the recent years. It is undeniable that the concepts underlying Transfer Optimization are formulated on solid grounds. However, evidences observed in recent contributions confirm that there are critical aspects that are not properly addressed to date. This short communication aims to engage the readership around a reflection on these issues, and to provide rationale why they remain unsolved. Specifically, we emphasize on three critical points of Evolutionary Multitasking Optimization: i) the plausibility and practical applicability of this paradigm; ii) the novelty of some proposed multitasking methods; and iii) the methodologies used for evaluating newly proposed multitasking algorithms. As a result of this research, we conclude that some important efforts should be directed by the community in order to keep the future of this promising field on the right track. Our ultimate purpose is to unveil gaps in the current literature, so that prospective works can attempt to fix these gaps, avoiding to stumble on the same stones and eventually achieve valuable advances in the area.