The application of large deep learning models in weather forecasting has led to significant advancements in the field, including higher-resolution forecasting and extended prediction periods exemplified by models such as Pangu and Fuxi. Despite these successes, previous research has largely been characterized by the neglect of extreme weather events, and the availability of datasets specifically curated for such events remains limited. Given the critical importance of accurately forecasting extreme weather, this study introduces a comprehensive dataset that incorporates high-resolution extreme weather cases derived from the High-Resolution Rapid Refresh (HRRR) data, a 3-km real-time dataset provided by NOAA. We also evaluate the current state-of-the-art deep learning models and Numerical Weather Prediction (NWP) systems on HR-Extreme, and provide a improved baseline deep learning model called HR-Heim which has superior performance on both general loss and HR-Extreme compared to others. Our results reveal that the errors of extreme weather cases are significantly larger than overall forecast error, highlighting them as an crucial source of loss in weather prediction. These findings underscore the necessity for future research to focus on improving the accuracy of extreme weather forecasts to enhance their practical utility

Quadrupedal locomotion via Reinforcement Learning (RL) is commonly addressed using the teacher-student paradigm, where a privileged teacher guides a proprioceptive student policy. However, key challenges such as representation misalignment between privileged teacher and proprioceptive-only student, covariate shift due to behavioral cloning, and lack of deployable adaptation; lead to poor generalization in real-world scenarios. We propose Teacher-Aligned Representations via Contrastive Learning (TAR), a framework that leverages privileged information with self-supervised contrastive learning to bridge this gap. By aligning representations to a privileged teacher in simulation via contrastive objectives, our student policy learns structured latent spaces and exhibits robust generalization to Out-of-Distribution (OOD) scenarios, surpassing the fully privileged “Teacher”. Results showed accelerated training by 2× compared to state-of-the-art baselines to achieve peak performance. OOD scenarios showed better generalization by 40% on average compared to existing methods. Moreover, TAR transitions seamlessly into learning during deployment without requiring privileged states, setting a new benchmark in sample-efficient, adaptive locomotion and enabling continual fine-tuning in real-world scenarios. Open-source code and videos are available at https://amrmousa.com/TARLoco/.

Ring attractors, mathematical models inspired by neural circuit dynamics, provide a biologically plausible mechanism to improve learning speed and accuracy in Reinforcement Learning (RL). Serving as specialized brain-inspired structures that encode spatial information and uncertainty, ring attractors explicitly encode the action space, facilitate the organization of neural activity, and enable the distribution of spatial representations across the neural network in the context of Deep Reinforcement Learning (DRL). These structures also provide temporal filtering that stabilizes action selection during exploration, for example, by preserving the continuity between rotation angles in robotic control or adjacency between tactical moves in game-like environments. The application of ring attractors in the action selection process involves mapping actions to specific locations on the ring and decoding the selected action based on neural activity. We investigate the application of ring attractors by both building an exogenous model and integrating them as part of DRL agents. Our approach significantly improves state-of-the-art performance on the Atari 100k benchmark, achieving a 53\% increase in performance over selected baselines.

Synthetic data has a key role to play in data sharing by statistical agencies and other generators of statistical data products. Generative Adversarial Networks (GANs), typically applied to image synthesis, are also a promising method for tabular data synthesis. However, there are unique challenges in tabular data compared to images, eg tabular data may contain both continuous and discrete variables and conditional sampling, and, critically, the data should possess high utility and low disclosure risk (the risk of re-identifying a population unit or learning something new about them), providing an opportunity for multi-objective (MO) optimization. Inspired by MO GANs for images, this paper proposes a smart MO evolutionary conditional tabular GAN (SMOE-CTGAN). This approach models conditional synthetic data by applying conditional vectors in training, and uses concepts from MO optimisation to balance disclosure risk against utility. Our results indicate that SMOE-CTGAN is able to discover synthetic datasets with different risk and utility levels for multiple national census datasets. We also find a sweet spot in the early stage of training where a competitive utility and extremely low risk are achieved, by using an Improvement Score. The full code can be downloaded from github1.

The number of home caretakers is rising rapidly due to an increasing number of elderly people, recent pandemics, and the advancement of home health care facilities. Wearable medical devices and the Internet of Medical Things (IoMT) help health care managers monitor patients in real-time and provide remote medical care. This reduces home visits and helps Home Health Care (HHC) companies plan their resources. The paper addresses the HHC planning problem of allocating the optimal number of experts to patients while minimising the delay in visiting the patient, matching medical expertise with patient needs, and identifying the patient’s visit sequence. To tackle this, a new mixed-integer mathematical problem is proposed to reduce the total visit time for patients. This paper makes three key contributions towards tackling this plan, including (i) providing a formal definition of the problem and putting it in context with related work, (ii) proposing multiple problem instances varying in complexity, and (iii) an initial analysis of several heuristics and an exact solver (CPLEX) on these problem instances. The results indicated that the application of computational intelligence combined with IoMT can reduce patient visitation time significantly in a daily plan and therefore lead to 3.7 percent improved care for HHC patients.

Abstract In order to trust the predictions of a machine learning algorithm, it is necessary to understand the factors that contribute to those predictions. In the case of probabilistic and uncertainty-aware models, it is necessary to understand not only the reasons for the predictions themselves, but also the reasons for the model’s level of confidence in those predictions. In this paper, we show how existing methods in explainability can be extended to uncertainty-aware models and how such extensions can be used to understand the sources of uncertainty in a model’s predictive distribution. In particular, by adapting permutation feature importance, partial dependence plots, and individual conditional expectation plots, we demonstrate that novel insights into model behaviour may be obtained and that these methods can be used to measure the impact of features on both the entropy of the predictive distribution and the log-likelihood of the ground truth labels under that distribution. With experiments using both synthetic and real-world data, we demonstrate the utility of these approaches to understand both the sources of uncertainty and their impact on model performance.

Evolutionary multi-objective optimization algorithms (EMOAs) typically assume that all objectives that are relevant to the decision-maker (DM) are optimized by the EMOA. In some scenarios, however, there are irrelevant objectives that are optimized by the EMOA but ignored by the DM, as well as, hidden objectives that the DM considers when judging the utility of solutions but are not optimized. This discrepancy between the EMOA and the DM’s preferences may impede the search for the most-preferred solution and waste resources evaluating irrelevant objectives. Research on objective reduction has focused so far on the structure of the problem and correlations between objectives and neglected the role of the DM. We formally define here the concepts of irrelevant and hidden objectives and propose methods for detecting them, based on uni-variate feature selection and recursive feature elimination, that use the preferences already elicited when a DM interacts with a ranking-based interactive EMOA (iEMOA). We incorporate the detection methods into an iEMOA capable of dynamically switching the objectives being optimized. Our experiments show that this approach can efficiently identify which objectives are relevant to the DM and reduce the number of objectives being optimized, while keeping and often improving the utility, according to the DM, of the best solution found.

Facility location problems (FLPs) are one of the most studied problem classes in supply chain management. However, despite the high number of research outputs, complex FLPs with large decision spaces and multi-objective formulations remain hard to solve. In this paper we introduce a multi-objective mathematical model for the FLP in personalised medicine, and apply a multi-stage algorithmic approach to solve it. In this case, the supply chain is circular and follows an on-demand and batch specific approach where the patient is also the donor. We solve the problem in a multi-stage manner, each stage optimising a sub-space of the larger decision space. In each stage we free up more decision variables to optimise, until eventually all decision variables defining the complete problem are made available for optimisation. A variant of the NSGA-II algorithm is used as solution method to solve both the complete problem and the different problem stages. Our results suggest that the multi-stage approach is able to find better solutions when compared to an approach that is given an equivalent number of evaluations but optimises the complete problem at once.

Quantum and quantum-inspired optimisation algorithms are designed to solve problems represented in binary, quadratic and unconstrained form. Combinatorial optimisation problems are therefore often formulated as Quadratic Unconstrained Binary Optimisation Problems (QUBO) to solve them with these algorithms. Moreover, these QUBO solvers are often implemented using specialised hardware to achieve enormous speedups, e.g. Fujitsu's Digital Annealer (DA) and D-Wave's Quantum Annealer. However, these are single-objective solvers, while many real-world problems feature multiple conflicting objectives. Thus, a common practice when using these QUBO solvers is to scalarise such multi-objective problems into a sequence of single-objective problems. Due to design trade-offs of these solvers, formulating each scalarisation may require more time than finding a local optimum. We present the first attempt to extend the algorithm supporting a commercial QUBO solver as a multi-objective solver that is not based on scalarisation. The proposed multi-objective DA algorithm is validated on the bi-objective Quadratic Assignment Problem. We observe that algorithm performance significantly depends on the archiving strategy adopted, and that combining DA with non-scalarisation methods to optimise multiple objectives outperforms the current scalarised version of the DA in terms of final solution quality.

We consider a type of constrained optimization problem, where the violation of a constraint leads to an irrevocable loss, such as breakage of a valuable experimental resource/platform or loss of human life. Such problems are referred to as safe optimization problems (SafeOPs). While SafeOPs have received attention in the machine learning community in recent years, there was little interest in the evolutionary computation (EC) community despite some early attempts between 2009 and 2011. Moreover, there is a lack of acceptable guidelines on how to benchmark different algorithms for SafeOPs, an area where the EC community has significant experience in. Driven by the need for more eficient algorithms and benchmark guidelines for SafeOPs, the objective of this paper is to reignite the interest of the EC community in this problem class. To achieve this we (i) provide a formal definition of SafeOPs and contrast it to other types of optimization problems that the EC community is familiar with, (ii) investigate the impact of key SafeOP parameters on the performance of selected safe optimization algorithms, (iii) benchmark EC against state-of-the-art safe optimization algorithms from the machine learning community, and (iv) provide an open-source Python framework to replicate and extend our work.

We consider a new class of expensive, resource-constrained optimization problems (here arising from molecular discovery) where costs are associated with the experiments (or evaluations) to be carried out during the optimization process. In the molecular discovery problem, candidate compounds to be optimized must be synthesized in an iterative process that starts from a set of purchasable items and builds up to larger molecules. To produce target molecules, their required resources are either used from already-synthesized items in storage or produced themselves on-demand at an additional cost. Any remaining resources from the production process are stored for reuse for the next evaluations. We model these resource dependencies with a directed acyclic production graph describing the development process from granular purchasable items to evaluable target compounds. Moreover, we develop several resource-eficient algorithms to address this problem. In particular, we develop resource-aware variants of Random Search heuristics and of Bayesian Optimization and analyze their performance in terms of anytime behavior. The experimental results were obtained from a real-world molecular optimization problem. Our results suggest that algorithms that encourage exploitation by reusing existing resources achieve satisfactory results while using fewer resources overall.

Visualizing the search behavior of a series of points or populations in their native domain is critical in understanding biases and attractors in an optimization process. Distance-based many-objective optimization test problems have been developed to facilitate visualization of search behavior in a 2-D design space with arbitrarily many objective functions. Previous works have proposed a few commonly seen problem characteristics into this problem framework, such as the definition of disconnected Pareto sets and dominance resistant regions of the design space. The authors’ previous work has advanced this research further by providing a problem generator to automatically create user-defined problem instances featuring any combination of these problem features as well as newly introduced ones, such as landscape discontinuities, varying objective ranges, and neutrality. This work makes a number of additional contributions including the proposal of an enhanced, open-source feature-rich problem generator that can create user-defined problem instances exhibiting a range of problem features—some of which are newly introduced here or form extensions of existing features. A comprehensive validation of the problem generator is also provided using popular multiobjective optimization algorithms, and some problem generator settings to create instances exhibiting different challenges for an optimizer are identified.

Comprehensive benchmarking of clustering algorithms is rendered difficult by two key factors: 1) the elusiveness of a unique mathematical definition of this unsupervised learning approach and 2) dependencies between the generating models or clustering criteria adopted by some clustering algorithms and indices for internal cluster validation. Consequently, there is no consensus regarding the best practice for rigorous benchmarking, and whether this is possible at all outside the context of a given application. Here, we argue that synthetic datasets must continue to play an important role in the evaluation of clustering algorithms, but that this necessitates constructing benchmarks that appropriately cover the diverse set of properties that impact clustering algorithm performance. Through our framework, HAWKS, we demonstrate the important role evolutionary algorithms play to support flexible generation of such benchmarks, allowing simple modification and extension. We illustrate two possible uses of our framework: 1) the evolution of benchmark data consistent with a set of hand-derived properties and 2) the generation of datasets that tease out performance differences between a given pair of algorithms. Our work has implications for the design of clustering benchmarks that sufficiently challenge a broad range of algorithms, and for furthering insight into the strengths and weaknesses of specific approaches.

In optimiser analysis and design it is informative to visualise how a search point/population moves through the design space over time. Visualisable distance-based many-objective optimisation problems have been developed whose design space is in two-dimensions with arbitrarily many objective dimensions. Previous work has shown how disconnected Pareto sets may be formed, how problems can be projected to and from arbitrarily many design dimensions, and how dominance resistant regions of design space may be defined. Most recently, a test suite has been proposed using distances to lines rather than points. However, active use of visualisable problems has been limited. This may be because the type of problem characteristics available has been relatively limited compared to many practical problems (and non-visualisable problem suites). Here we introduce the mechanisms required to embed several widely seen problem characteristics in the existing problem framework. These include variable density of solutions in objective space, landscape discontinuities, varying objective ranges, neutrality, and non-identical disconnected Pareto set regions. Furthermore, we provide an automatic problem generator (as opposed to hand-tuned problem definitions). The flexibility of the problem generator is demonstrated by analysing the performance of popular optimisers on a range of sampled instances.

Synthetic datasets play an important role in evaluating clustering algorithms, as they can help shed light on consistent biases, strengths, and weaknesses of particular techniques, thereby supporting sound conclusions. Despite this, there is a surprisingly small set of established clustering benchmark data, and many of these are currently handcrafted. Even then, their difficulty is typically not quantified or considered, limiting the ability to interpret algorithmic performance on these datasets. Here, we introduce HAWKS, a new data generator that uses an evolutionary algorithm to evolve cluster structure of a synthetic data set. We demonstrate how such an approach can be used to produce datasets of a pre-specified difficulty, to trade off different aspects of problem difficulty, and how these interventions directly translate into changes in the clustering performance of established algorithms.

We consider multiobjective optimization problems where objective functions have different (or heterogeneous) evaluation times or latencies. This is of great relevance for (computationally) expensive multiobjective optimization as there is no reason to assume that all objective functions should take an equal amount of time to be evaluated (particularly when objectives are evaluated separately). To cope with such problems, we propose a variation of the Kriging-assisted reference vector guided evolutionary algorithm (K-RVEA) called heterogeneous K-RVEA (short HK-RVEA). This algorithm is a merger of two main concepts designed to account for different latencies: A single-objective evolutionary algorithm for selecting training data to train surrogates and K-RVEA's approach for updating the surrogates. HK-RVEA is validated on a set of biobjective benchmark problems varying in terms of latencies and correlations between the objectives. The results are also compared to those obtained by previously proposed strategies for such problems, which were embedded in a non-surrogate-assisted evolutionary algorithm. Our experimental study shows that, under certain conditions, such as short latencies between the two objectives, HK-RVEA can outperform the existing strategies as well as an optimizer operating in an environment without latencies.

A key consideration in developing optimization approaches for data clustering is choice of a suitable encoding. Existing encodings strike different trade-offs between model and search complexity, limiting the applicability to data sets with particular properties or to problems of moderate size. Recent research has introduced an additional hyperparameter to directly govern the encoding granularity in the multi-objective clustering algorithm MOCK. Here, we investigate adapting this important hyperparameter during run-time. In particular, we consider a number of different trigger mechanisms to control the timing of changes to this hyperparameter and strategies to rapidly explore the newly "opened" search space resulting from this change. Experimental results illustrate distinct performance differences between the approaches tested, which can be explained in light of the relative importance of initialization, crossover and mutation in MOCK. The most successful strategies meet the clustering performance achieved for an optimal (a priori) setting of the hyperparameter, at a ~40% reduction of computational expense.

Substantial research has been carried out to explain the effects of economic variables on obesity, typically considering only a few factors at a time, using parametric linear regression models. Recent studies have made a significant contribution by examining economic factors affecting body weight using the Behavioral Risk Factor Surveillance System data with 27 state-level variables for a period of 20 years (1990-2010). As elsewhere, the authors solely focus on individual effects of potential drivers of obesity than critical interactions among the drivers. We take some steps to extend the literature and gain a deeper understanding of the drivers of obesity. We employ state-of-the-art data mining techniques to uncover critical interactions that may exist among drivers of obesity in a data-driven manner. The state-of-the-art techniques reveal several complex interactions among economic and behavioral factors that contribute to the rise of obesity. Lower levels of obesity, measured by a body mass index (BMI), belong to female individuals who exercise outside work, enjoy higher levels of education and drink less alcohol. The highest level of obesity, in contrast, belongs to those who fail to exercise outside work, smoke regularly, consume more alcohol and come from lower income groups. These and other complementary results suggest that it is the joint complex interactions among various behavioral and economic factors that gives rise to obesity or lowers it; it is not simply the presence or absence of individual factors.

Real-world optimization problems are often subject to several constraints which are expensive to evaluate in terms of cost or time. Although a lot of effort is devoted to make use of surrogate models for expensive optimization tasks, not many strong surrogate-assisted algorithms can address the challenging constrained problems. Efficient Global Optimization (EGO) is a Kriging-based surrogate-assisted algorithm. It was originally proposed to address unconstrained problems and later was modified to solve constrained problems. However, these type of algorithms still suffer from several issues, mainly: (1) early stagnation, (2) problems with multiple active constraints and (3) frequent crashes. In this work, we introduce a new EGO-based algorithm which tries to overcome these common issues with Kriging optimization algorithms. We apply the proposed algorithm on problems with dimension d ≤ 4 from the G-function suite [16] and on an airfoil shape example.

This study considers an actual real-world problem encountered by ARM, the world's leading semiconductor intellectual property (IP) supplier, concerning the multi-year assignment of weeks-long computationally intensive projects (executable pieces of code) across a number of capacity-limited clusters. The quality of a projects-to-cluster assignment is measured in terms of several metrics such as the even utilization of clusters, being able to realize all projects, and spreading projects of different research groups evenly across the clusters. The first (theoretical) contribution of this work is to motivate and formally define this novel application and put it in context with related literature. The second (experimental) contribution of this work is about gaining an understanding about the problem and performing an initial investigation on how different algorithm types (random search, an EMOA, and greedy search) fare on the problem. Our study revealed that the problem has many infeasible solutions and is challenging to optimize especially for long planning horizons (more than 3 years). While the EMOA is able to outperform random and greedy search (and also the current approach used at ARM) in terms of solution quality discovered, greedy search was the computationally most efficient approach and suitable for short term planning horizons (up to 1 year).

Market mechanisms are a means by which resources in contention can be allocated between contending parties, both in human economies and those populated by software agents. Designing such mechanisms has traditionally been carried out by hand, and more recently by automation. Assessing these mechanisms typically involves them being evaluated with respect to multiple conflicting objectives, which can often be nonlinear, noisy, and expensive to compute. For typical performance objectives, it is known that designed mechanisms often fall short on being optimal across all objectives simultaneously. However, in all previous automated approaches, either only a single objective is considered, or else the multiple performance objectives are combined into a single objective. In this paper we do not aggregate objectives, instead considering a direct, novel application of multi-objective evolutionary algorithms (MOEAs) to the problem of automated mechanism design. This allows the automatic discovery of trade-offs that such objectives impose on mechanisms. We pose the problem of mechanism design, specifically for the class of linear redistribution mechanisms, as a naturally existing multi-objective optimisation problem. We apply a modified version of NSGA-II in order to design mechanisms within this class, given economically relevant objectives such as welfare and fairness. This application of NSGA-II exposes tradeoffs between objectives, revealing relationships between them that were otherwise unknown for this mechanism class. The understanding of the trade-off gained from the application of MOEAs can thus help practitioners with an insightful application of discovered mechanisms in their respective real/artificial markets.

We consider optimization problems where the set of solutions available for evaluation at any given time t during optimization is some subset of the feasible space. This model is appropriate to describe many closed-loop optimization settings (i.e., where physical processes or experiments are used to evaluate solutions) where, due to resource limitations, it may be impossible to evaluate particular solutions at particular times (despite the solutions being part of the feasible space). We call the constraints determining which solutions are non-evaluable ephemeral resource constraints (ERCs). In this paper, we investigate two specific types of ERC: one encodes periodic resource availabilities, the other models commitment constraints that make the evaluable part of the space a function of earlier evaluations conducted. In an experimental study, both types of constraint are seen to impact the performance of an evolutionary algorithm significantly. To deal with the effects of the ERCs, we propose and test five different constraint-handling policies (adapted from those used to handle standard constraints), using a number of different test functions including a fitness landscape from a real closed-loop problem. We show that knowing information about the type of resource constraint in advance may be sufficient to select an effective policy for dealing with it, even when advance knowledge of the fitness landscape is limited.

We consider an optimization scenario in which resources are required in the evaluation process of candidate solutions. The challenge we are focussing on is that certain resources have to be committed to for some period of time whenever they are used by an optimizer. This has the effect that certain solutions may be temporarily non-evaluable during the optimization. Previous analysis revealed that evolutionary algorithms (EAs) can be effective against this resourcing issue when augmented with static strategies for dealing with non-evaluable solutions, such as repairing, waiting, or penalty methods. Moreover, it is possible to select a suitable strategy for resource-constrained problems offline if the resourcing issue is known in advance. In this paper we demonstrate that an EA that uses a reinforcement learning (RL) agent, here Sarsa(λ), to learn offline when to switch between static strategies, can be more effective than any of the static strategies themselves. We also show that learning the same task as the RL agent but online using an adaptive strategy selection method, here D-MAB, is not as effective; nevertheless, online learning is an alternative to static strategies.