ABSTRACT. Instrument recognition in polyphonic audio signals is a very challenging classification task. It helps to improve related application scenarios, like music transcription and recommendation, organization of large music collections, or analysis of historical trends and properties of musical styles. Recently, the classification performance could be improved by the integration of deep convolutional neural networks. However, in to date published studies, the network architectures and parameter settings were usually adopted from image recognition tasks and manually adjusted, without a systematic optimization. In this paper, we show how two different neural architecture search strategies can be successfully applied for the prediction of nine instrument classes, significantly outperforming the classification performance of three fixed baseline architectures from previous works. Although high computing efforts for model optimization are required, the training of the final architecture is done only once for later prediction of instruments in a possibly unlimited number of musical tracks.

ABSTRACT. We propose a methodology for music generation that makes use of Ant Colony Optimization (ACO) algorithms on multilayer graphs. In our methodology we first define a new multilayer graph model of music that contains several voices musical works. Then, we adapt ACO algorithms to allow multiple ant colonies to generate solutions on each layer while interacting with each other. This methodology is illustrated with some example configurations that show how music emerge as a result of the interaction of different simultaneous ant colony optimization instances.

ABSTRACT. Despite their impressive offline results, deep learning models for symbolic music generation are not widely used in live performances due to a deficit of musically meaningful control parameters and a lack of structured musical form in their outputs. To address these issues we introduce LooperGP, a method for steering a Transformer-XL model towards generating loopable musical phrases of a specified number of bars and time signature, enabling a tool for live coding performances. We show that by training LooperGP on a dataset of 93,681 musical loops extracted from the DadaGP datset, we are able to steer its generative output towards generating 3x as many loopable phrases as our baseline. In a subjective listening test conducted by 31 participants, LooperGP loops achieved positive median ratings in originality, musical coherence and loop smoothness, demonstrating its potential as a performance tool.

ABSTRACT. Thanks to the efforts and cooperation of the international community, nowadays it is possible to analyze astronomical data captured by the observatories and telescopes of major space agencies around the world from a personal computer. The development of virtual observatory technology (VO), and the standardization of the formats it uses, allow professional and amateur astronomers to access astronomical data and images through internet with relative ease. Immersed in this environment of global accessibility, this article presents an astronomical data-driven unsupervised music composition system based on \textit{Deep Learning}, aimed at offering an automatic and objective review on the classical topic of the \textit{Harmonies of the Spheres}. The system explores the \textit{MILES} stellar library from the Spanish Virtual Observatory (SVO) using a variational autoencoder architecture to cross-match its stellar spectra via \textit{Pitch-Class Set Theory} with a music score generated by a LSTM with attention neural network in the style of late-renaissance music.

ABSTRACT. This study addresses the two-echelon electric vehicle routing problem with simultaneous pickup and deliveries. In a two-echelon distribution network, large vehicles transport goods from central warehouses to satellites, while smaller and environmentally friendly vehicles distribute goods from these satellites to final customers. The considered problem also includes simultaneous pickup and delivery constraints that usually arise as a reverse logistics practice. A MILP model is developed and solved for small-sized problem instances using CPLEX. Since the tackled problem becomes rather complex because of the multi-tier structure and constraints, solving even small-sized instances using CPLEX requires very long computation times. Therefore, the application of a self-adaptive variant of the hybrid metaheuristic Construct, Merge, Solve & Adapt is proposed. In the context of problem instances too large for the application of CPLEX, our algorithm is compared to probabilistic versions of two well-known constructive heuristics. The numerical results show that our algorithm outperforms CPLEX in the context of rather small problem instances. Moreover, it is shown to outperform the heuristic algorithms when larger problem instances are concerned.

ABSTRACT. Our work addresses a real-world freight transportation problem with a broad set of characteristics. We build upon the classical work of Ropke and Pisinger [] and propose an effective realization of the adaptive large neighborhood search (ALNS) with constant time complexity for a large portion of frequent steps in insertion and removal heuristics at the cost of additional pre-calculations. Our minimization process handles different objectives with cost models of heterogeneous vehicles. We demonstrate the generic applicability of the proposed solver on various vehicle routing problems. With the help of the standard Li & Lim benchmarks [] for pickup and delivery with time windows, we show its capabilities compared to the best-found solutions and the original ALNS. Experiments on real-world delivery routing problems provide a comparison with the company's original implementation in OR-Tools [], where we achieve significant cost savings, faster runtime, and memory savings in the order of magnitude. Performance on large-scale real-world instances with more than 300 vehicles and 1,200 pickup and delivery requests is also presented, while computing results in less than an hour.

ABSTRACT. We propose a multilevel optimization algorithm (MLO) for solving large scale instances of the Multi-Period Battery Swapping Station Location Problem (MBSSLP), i.e., a problem for deciding the placement of battery swapping stations in an urban area. MLO generates a solution to an MBSSLP instance in three steps. First the problem size is iteratively reduced by coarsening. Then, a solution to the coarsest problem instance is determined, and finally the obtained solution is projected to more fine grained problem instances in reverse order until a solution to the original problem instance is obtained. We test our approach on benchmark instances with up to 10000 areas for placing stations and 100000 user trips. We compare MLO to solving a mixed integer linear program (MILP) in a direct way as well as solving the instances with a construction heuristic (CH). Results show that MLO scales substantially better for such large instances than the MILP or the CH.

ABSTRACT. Fitness functions fail to differentiate between different solutions with the same fitness, and this lack of ability to distinguish between solutions can have a detrimental effect on the search process. We investigate, for the Travelling Salesman Problem (TSP), the impact of using a hash function to differentiate solutions during the search process. Whereas this work is not intended to improve the state-of-the-art of the TSP solvers, it nevertheless reveals a positive effect when the hash function is used.

ABSTRACT. In this study, we developed an algorithm that provides automatic protection against swarm drones by using directional jammers in anti-drone systems. Directional jammers are a special type of jammers that can be rotated to a certain angle and do jamming around only that angle. This feature is useful for jamming particular targets and not jamming areas where it is not desired. We worked on a specialized version of the threat evaluation and weapon allocation (TEWA) problem, in which weapons (jammers for this problem) should be assigned to angles to cover threats at a maximum rate by satisfying their priorities. In this problem, it is aimed at keeping the threat within jamming signals at the maximum rate by turning the directional jammers to appropriate angles, taking into account the threat priorities. We have presented an algorithm that solves this problem by meeting the physical constraints of the jammers and tactical constraints defined by the user. The solutions created include: using as few jammers as possible, minimizing the angle changes jammers make in each new plan, prioritizing threats according to their characteristics (type, direction, speed, and distance), and preventing jammers from returning to the physical constraints defined for them. We solved the threat evaluation problem with the help of genetic programming and the jammer angle assignment problem by transforming it into an integer linear programming (ILP) formulation. We also handled physical constraints unsuitable for ILP formulation with post-processing. Since there are few studies directly dealing with this version of the problem, we compared our study with the study that was claimed to be the first to solve this particular version of this problem. Furthermore, we compared our study with the different versions of the algorithm we created. Experiments have shown that threat coverage percentage is vastly increased, achieving this without a significant drop in problem-solving speed.

ABSTRACT. Empirical data plays an important role in evolutionary computation research. To make better use of the available data, ontologies have been proposed in the literature to organize their storage in a structured way. However, the full potential of these formal methods to capture our domain knowledge has yet to be demonstrated. In this work, we evaluate a performance prediction model built on top of the extension of the recently proposed OPTION ontology. More specifically, we first extend the OPTION ontology with the vocabulary needed to represent modular black-box optimization algorithms. Then, we use the extended OPTION ontology, to create knowledge graphs with fixed-budget performance data for two modular algorithm frameworks, modCMA, and modDE, for the 24 noiseless BBOB benchmark functions. We build the performance prediction model using a knowledge graph embedding-based methodology. Using a number of different evaluation scenarios, we show that a triple classification approach, a fairly standard predictive modeling task in the context of knowledge graphs, can correctly predict whether a given algorithm instance will be able to achieve a certain target precision for a given problem instance. This approach requires feature representation of algorithms and problems. While the latter is already well developed, we hope that our work will motivate the community to collaborate on appropriate algorithm representations.

ABSTRACT. Aiming to respond to a real-world complex problem, optimizing an electricity mix production, MixSimulator has been developed. This paper compares two approaches: the classic method and the multi-agent system based method (MAS). Each agent performs various functions such as producing electricity and updating the availability (power plants), predicting the oncoming demand, and handling all the information to provide optimized planning of the production. It takes into account technological, economic, and environmental constraints. Black-Box optimizers are used to generate solutions. The results show that the multi-agent system based method outperforms the classic one thanks to its ability to react to events and provide dynamic schedule.

ABSTRACT. S-boxes are an important primitive that help cryptographic algorithms to be resilient against various attacks. The resilience against specific attacks can be connected with a certain property of an S-box, and the better the property value, the more secure the algorithm. One example of such a property is called boomerang uniformity, which helps to be resilient against boomerang attacks. How to construct S-boxes with good boomerang uniformity is not always clear. There are algebraic techniques that can result in good boomerang uniformity, but the results are still rare.

In this work, we explore the evolution of S-boxes with good values of boomerang uniformity. We consider three different encodings and five S-box sizes. For sizes $4\times 4$ and $5\times 5$, we manage to obtain optimal solutions. For $6\times 6$, we obtain optimal boomerang uniformity for the non-APN function. For larger sizes, the results indicate the problem to be very difficult (even more difficult than evolving differential uniformity, which can be considered a well-researched problem).

ABSTRACT. Non-cryptographic (NC) hash functions are the core part of many networking and security applications such as traffic flow monitoring and deep packet inspection. For these applications, speed is more important than strong cryptographic properties. In Terabit Ethernet networks, the speed of the hash functions can have a significant impact on the overall performance of the system when it is required to process the packets at a line rate. Hence improving the speed of hash functions can have a significant impact on the overall performance of such architectures. Designing a good hash function is a challenging task because of the highly non-linear and complex relationship between input and output variables. Techniques based on Evolutionary Computation (EC) excel in addressing such challenges. In this paper, we propose novel fast noncryptographic hash functions using genetic programming and we call the resulting hash functions the GPNCH (Genetic Programming-based Non- Cryptographic Hash) family. We choose to employ avalanche metrics as a fitness function because the networking and security applications we consider require hash functions to be uniform and independent. We evaluate the performance of GPNCH functions on FPGA and compare the delay, throughput, and resource occupation with the state-of-the-art NC hash functions that satisfy the avalanche criteria. We show that GPNCH functions outperform the other algorithms in terms of latency, operating frequency, and throughput at the modest cost of hardware resources.

ABSTRACT. Feature selection is important nowadays due to many real-world datasets usually having a large number of features. Evolutionary multi-objective optimisation algorithms have been successfully used for feature selection which usually has two conflicting objectives, i.e., maximising the classification accuracy and minimising the number of selected features. However, most of the existing evolutionary multi-objective feature selection algorithms tend to address feature selection tasks independently, even when these feature selection tasks are related. Multi-task optimisation, which aims to improve the performance of multiple tasks by sharing common knowledge among them, has been used in many areas. However, there is not much work on utilising multi-task optimisation for feature selection. In this work, we develop a new multi-task multi-objective feature selection algorithm. This algorithm aims to address multiple related feature selection tasks simultaneously and facilitate knowledge capturing and transferring among the related tasks. Furthermore, a method is developed for transferring knowledge between related feature selection tasks having different features. This method can avoid transferring information between the unique features of tasks by transforming the probability models of them. We compare the proposed algorithm with the single-task multi-objective feature selection algorithm on seven sets of related feature selection tasks. Experimental results show that the proposed algorithm achieves better classification performance than the single-task algorithm with the help of knowledge transferring among related feature selection tasks. Further analysis shows that the features selected by our proposed algorithm can be more relevant to the classification tasks.

ABSTRACT. Feature selection is becoming an essential part of machine learning pipelines, including the ones generated by recent AutoML tools. In case of datasets with epistatic interactions between the features, like many datasets from the bioinformatics domain, feature selection may even become crucial. A recent method called SLUG has outperformed the state-of-the-art algorithms for feature selection on a large set of epistatic noisy datasets. SLUG uses genetic programming (GP) as a classifier (learner), nested inside a genetic algorithm (GA) that performs feature selection (wrapper). In this work, we pair GA with different learners, in an attempt to match the results of SLUG with less computational effort. We also propose a new feedback mechanism between the learner and the wrapper to improve the convergence towards the key features. Although we do not match the results of SLUG, we demonstrate the positive effect of the feedback mechanism, motivating additional research in this area to further improve SLUG and other existing feature selection methods.

ABSTRACT. Developing effective and adaptive robotic arm controllers is crucial for many industries, e.g. manufacturing. Traditional pre-programmed controllers cannot adapt to changing environments. This study investigates how neuroevolution can be used to develop robotic arm controllers and addresses key gaps in the existing literature, such as incorporating expert demonstrations and analyzing the robustness of evolved controllers. In addition to addressing these questions, this work compares different controller architectures and training algorithms. A evolutionary supervisor neural network approach is also proposed to solve the pick-and-place task. The proposed method achieves a high successful completion rate, 927 out of 1000 trials.

ABSTRACT. Deep learning is an effective and efficient method for image segmentation. Several neural network designs have been investigated, a notable example being the U-Net which has outperformed other segmentation models in different challenges. The Spatial Attention U-Net is a variation of the U-Net, which utilizes DropBlock and an attention block in addition to the typical U-Net convolutional blocks, which boosted the accuracy of the U-Net and reduced over-fitting. Optimising neural networks is costly, time-consuming and often requires expert guidance to determine the best mix of hyper-parameters for a particular problem. We demonstrate that grammatical evolution (GE) can be used to create U-Net and SA-UNet architectures and optimise its choice of hyper- parameters. Our results show improved performance over state-of-the-art models on the Retinal Blood Vessel problem, increasing both AUC, from 0.978 to 0.979 and Accuracy, from 0.964 0.966, from the base models. Crucially, GE can achieve these improvements while finding a model which is 10 times smaller than the base models. A smaller model would enable its use in smart devices, such as smart phones, or in edge computing.

ABSTRACT. This paper focuses on explaining the results of Recommender Systems, that aim at suggesting, for a given user, the most accurate products, among a given set of available products, and modeling how different types of user activities, such as based on user interests in different categories of products, affect the results of the recommender system. It proposes an evolutionary approach to interests mix modeling that defines the relation between the characteristic of the user ratings and the composition of the list of the recommended products. Computational experiments, performed on some selected benchmarks derived from the MovieLens dataset, confirmed the accuracy and efficiency of the proposed approach.

ABSTRACT. The learning process of feedforward neural networks (FFNN), which includes determining the suitable weights for connections and biases, is one of the most challenging machine learning problems. The efficiency of the learning process has a significant impact on how well FFNNs work. Back-propagation (BP), a gradient descent-based method, is one of the most popular learning algorithms, however, it can get stuck in local optima. Differential evolution (DE), a popular population-based metaheuristic algorithm, is a reliable alternative for tackling challenging issues. In order to enhance the functionality of FFNNs, this paper suggests a centroid-base differential evolution with composite trial vector generation strategies and control parameters (Cen-CoDE). Our proposed algorithm employs a centroid-based strategy in three different ways to generate different trail vectors, simultaneously. Additionally, the objective function of our suggested algorithm is based on a classification error, whereas connection weights and biases are represented as a candidate solution. The performance of Cen-CoDE in comparison to other contemporary techniques is supported by experimental findings.

ABSTRACT. Feature Learning (FL) is key to well-performing machine learning models. However, the most popular FL methods lack interpretability, which is becoming a critical requirement of Machine Learning.

We propose to incorporate information from the problem domain in the structure of programs on top of the existing M3GP approach. This technique, named Domain-Knowledge M3GP, works by defining the possible feature transformations using a grammar through Grammar-Guided Genetic Programming. While requiring the user to specify the domain knowledge, this approach has the advantage of limiting the search space, excluding programs that make no sense to humans.

We extend this approach with the possibility of introducing complex, aggregating queries over historic data. This extension allows to expand the search space to include relevant programs that were not possible before.

We evaluate our methods on performance and interpretability in 6 use cases, showing promising results in both areas. We conclude that performance and interpretability of FL methods can benefit from domain-knowledge incorporation and aggregation, and give guidelines on when to use them.

ABSTRACT. Evolving LLVM IR is widely applicable, with LLVM Clang offering support for an increasing range of computer hardware and programming languages. Local search mutations are used to hill climb industry C code released to support geographic open standards: Open Location Code (OLC) from Google and Uber's Hexagonal Hierarchical Spatial Index (H3), giving up to two percent speed up on compiler optimised code.

ABSTRACT. An essential characteristic of brains in intelligent organisms is their spatial organization that different parts of a brain are responsible for solving different classes of problems. Inspired by biological brains, here we introduce Spatial Genetic Programming (SGP). In this GP paradigm, Linear Genetic Programming (LGP) programs, represented as graph nodes are spread in a 2D space. They are executed in an order determined by the network of interactions between program nodes: The execution of a program affects the selection of the next program in line. Different inputs to the proposed system trigger the execution of different locations in the GP model. As a proof of concept, performance and internal dynamics of SGP are tested for varying problems. We compare the performance of SGP with LGP and Tree GP and our results indicate that SGP manages to solve a diverse range of problems with different hyper-parameters. We also carry out an analysis of the spatial properties of SGP individuals.

ABSTRACT. Maintaining genetic diversity as a means to avoid premature convergence is critical in Genetic Programming. Several approaches have been proposed to achieve this, with some focusing on the mating phase from coupling dissimilar solutions to some form of self-adaptive selection mechanism. In nature, genetic diversity can be the consequence of many different factors, but when considering reproduction Sexual Selection can have an impact on promoting variety within a species. Specifically, Mate Choice often results in different selective pressures between sexes, which in turn may trigger evolutionary differences among them. Although some mechanisms of Sexual Selection have been applied to Genetic Programming in the past, the literature is scarce when it comes to Mate Choice. Recently, a way of modelling mating preferences by ideal mate representations was proposed, achieving good results when compared to a standard approach. These mating preferences evolve freely in a self-adaptive fashion, creating an evolutionary driving force of its own alongside fitness pressure. The inner mechanisms of this approach operate from personal choice, as each individual has its own representation of a perfect mate which affects the mate to be selected. In this paper, we compare this method against a random mate choice to assess whether there are advantages in evolving personal preferences. We conducted experiments using three symbolic regression problems and different mutation rates. The results show that self-adaptive mating preferences are able to create a more diverse set of solutions when compared to the traditional approach and a random mate approach (with statistically significant differences) and have a higher success rate in three of the six instances tested.

ABSTRACT. Evolutionary algorithms (EAs) are a family of optimization algorithms inspired by the Darwinian theory of evolution, and Genetic Algorithm (GA) is a popular technique among EAs. Similar to other EAs, common limitations of GAs have geometrical origins, like premature convergence, where the final population's convex hull might not include the global optimum. Population diversity maintenance is a central idea to tackle this problem but is often performed through methods that constantly diminish the search space's area. This work presents a self-adaptive approach, where the non-geometric crossover is strategically employed with geometric crossover to maintain diversity from a geometrical/topological perspective. To evaluate the performance of the proposed method, the experimental phase compares it against well-known diversity maintenance methods over well-known benchmarks. Experimental results clearly demonstrate the suitability of the proposed self-adaptive approach and the possibility of applying it to different types of crossover and EAs.

ABSTRACT. High-resolution simulations of particle-laden flows are computationally limited to a scale of thousands of particles due to the complex interactions between particles and fluid. Some approaches to increase the number of particles in such simulations require information about the fluid-induced force on a particle, which is a major challenge in this research area. In this paper, we present an approach to develop symbolic models for the fluid-induced force. We use a graph network as inductive bias to model the underlying pairwise particle interactions. The internal parts of the network are then replaced by symbolic models using a genetic programming algorithm. We include prior problem knowledge in our algorithm. The resulting equations show an accuracy in the same order of magnitude as state-of-the-art approaches for different benchmark datasets. They are interpretable and deliver important building blocks. Our approach is a promising alternative to ''black-box'' models from the literature.

ABSTRACT. This work proposes Adaptive Facilitated Mutation, a self-adaptive mutation method for Structured Grammatical Evolution (SGE), biologically inspired by the theory of facilitated variation. In SGE, the genotype of individuals contains a list for each non-terminal of the grammar that defines the search space. In our proposed mutation, each individual contains an array with a different, self-adaptive mutation rate for each non-terminal. We also propose Function Grouped Grammars, a grammar design procedure to enhance the benefits of the propose mutation. Experiments were conducted on three symbolic regression benchmarks using Probabilistic Structured Grammatical Evolution (PSGE), a variant of SGE. Results show our approach is similar or better when compared with the standard grammar and mutation.

ABSTRACT. Deep learning has recently established itself as a cluster of methods of choice for almost all classification tasks in music information retrieval. However, despite very good classification performance, it sometimes brings disadvantages including long training times and higher energy costs, lower interpretability of classification models, or an increased risk of overfitting when applied to small training sets due to a very large number of trainable parameters. In this paper, we investigate the combination of both deep and shallow algorithms for recognition of musical genres using a transfer learning approach. We train deep classification models once to predict harmonic, instrumental, and segment properties from datasets with respective annotations. Their predictions for another dataset with annotated genres are used as features for shallow classification methods. They can be trained over and again for different categories, and are particularly useful when the training sets are small, in a real world scenario when listeners define various musical categories selecting only a few prototype tracks. The experiments show the potential of the proposed approach for genre recognition. In particular, when combined with evolutionary feature selection which identifies the most relevant deep feature dimensions, the classification errors became significantly lower in almost all cases, compared to a baseline based on MFCCs or results reported in the previous work.

ABSTRACT. A system able to automatically transpose an audio recording would have many potential applications, from music production to hearing aid design. We present a deep learning approach to transpose an audio recording directly from the raw time domain signal. We train recurrent neural networks with samples of simple waveforms (sine, square, sawtooth, triangle) covering the range of possible frequencies in order to generate signals reflecting the desired frequency transposition. We examine our generated transpositions for each musical semitone step size up to the octave and compare our results against two popular pitch shifting algorithms. Although our approach is able to accurately transpose the frequencies in a signal, these signals suffer from a significant amount of added noise. This work represents exploratory steps towards the development of a general deep transposition model able to quickly transpose to any desired spectral mapping.

ABSTRACT. Graphic Design (GD) artefacts, like posters on the streets or book covers on store shelves, often compete with each other to be seen, catch attention and communicate effectively. Nevertheless, due to the democratisation of gd and because finding disruptive aesthetics might be time-consuming, graphic designers often follow existing trends, lacking disruptive and catchy visual features. Our system aims to assist the exploration of innovative GD aesthetics by employing a genetic algorithm to evolve content within two-dimensional pages. The system takes the form of an extension for Adobe InDesign, so both human designers and the machine can alternately collaborate in the creation process. In this paper, we propose a method to automatically evaluate the generated posters by assessing their dissimilarity to the output of an auto-encoder that was trained with a set of posters posted at typographicposters.com by graphic designers worldwide. The results suggest the viability of the evaluation method to recall large sets of images and therefore be used to compute an image dissimilarity degree. Furthermore, the proposed method can be used for evolving gd posters that can be deemed as new when compared to the training set.

ABSTRACT. Artificial intelligence has become a powerful creative tool that paradoxically often leaves little space for human creativity. For example, the aestheticized images are created in several seconds by submitting a few words to the imagine bot by Midjourney. What happens when neural networks meet materiality? This article uses practice-based research methodology to explore creative dialogue and possibilities between deep learning (DL) technology and physical volumetric form creation in ceramics. The artistic project discussed here juxtaposes the archaic craft method, such as ceramics, with innovative AI technology. [Project Name] is a series of case studies that explore DL possibilities for creating a tangible object guided by text prompt and a 3D model using a pipeline that mixes digital and analogue methods. The selected generated digital objects were manually altered and prepared for 3D printing in ceramics. Glazing happened by hand, often inspired by the AI-generated objects’ vertex colours. The artwork demonstrates embodied experience and transformation of DL model through artistic practice. The aim of the paper is to discover new meanings and discussion avenues that emerge by bringing together skilled manual and automated processes. In addition, we discuss the development of AI art and related sculpture works.

ABSTRACT. This paper introduces a two-fold methodology of creating fabric designs and patterns, using both traditional object detection and Deep Learning methodologies. The proposed methodology first augments a given partial sketch, which is taken as an input from the user. This sketch augmentation is performed through a combination of object de- tection, canvas quilting, and seamless tiling, to achieve a repeatable block of a pattern. This augmented pattern is then carried forward as an input to our variation of the pix2pix GAN, which outputs a styled and colored pattern using the sketch as a baseline. This design pipeline is an overall overhaul of the creative process of a textile designer, and is intended to provide assistance in the design of modern textiles in the industry by reducing the time from going to a sketch to a pattern in under a minute.

ABSTRACT. Automatic music genre classification systems are vital nowadays be-cause the traditional music genre classification process is mostly implemented without following a universal taxonomy and the traditional process for audio indexing is prone to error. Various techniques to implement an automatic music genre classification system can be found in the literature but the accuracy and efficiency of those systems are insufficient to make them useful for practical scenarios such as identifying songs by the music genre in radio broadcast monitoring systems. The main contribution of this research is to increase the accuracy and efficiency of current automatic music genre classification systems with a comprehensive analysis of correlations between the descriptive statistical features of audio signals and the music genres of songs. A greedy approach for music genre identification is also introduced to improve the accuracy and efficiency of music genre classification systems and to identify the music genre of complex songs that contain multiple music genres. The approach, proposed in this paper, reported 87.3% average accuracy for music genre classification on the GTZAN dataset over 10 music genres.

ABSTRACT. OSC-Qasm is a cross-platform, Python-based, OSC interface for executing Qasm code. It serves as a simple way to connect creative programming environments like Max (with The QAC Toolkit) and Pure Data with real quantum hardware, using the Open Sound Control protocol. In this paper, the authors introduce the context and meaning of developing a tool like this, and what it can offer to creative artists.

ABSTRACT. Simultaneously evolving morphologies (bodies) and controllers (brains) of robots for a specific task (objective) is challenging because it is not intuitively clear how the perfect morphology looks given a certain task and environment. Even when a morphology is found and fixed, learning its affiliated controller to perform well is a second step of elaborate experimentation. Naturally, the complexity is increasing with the number of objectives that the robot is supposed to solve. In this paper we are proposing an approach to tackle this problem in a multi-objective setup by showing an efficient way to achieve both goals, finding a good phenotype for a given task and environment and getting a controller that can be used to complete the task with this phenotype. To this end, we show how to solve the bi-objective optimization problem of rotating to search for a given target and then moving to its position. We find this to be a rather useful skill for both robots and animals and it is an analogy to the real-world need of search and hunt for food or other targets. To achieve it, we apply the NSGA-II algorithm for a multi-objective selection and to create a population of appropriate morphologies with plastic brains. Out of this population the most promising candidates are selected and two randomly initialized brains are evolved separately for the given tasks, that can then be used by a state machine to switch between the behaviors.

ABSTRACT. In recent years, much progress has been made in computer Go and most of the results have been obtained thanks to search algorithms (Monte Carlo Tree Search) and Deep Reinforcement Learning (DRL). In this paper, we propose to use and analyze the latest algorithms that use search and DRL (AlphaZero and Descent algorithms) to automatically learn to play an extended version of the game of Go with more than two players. We show that using search and DRL we were able to improve the level of play, even though there are more than two players.

ABSTRACT. Meta-heuristic stochastic optimization algorithms are predominantly used to solve complex real-world problems. Numerous new nature-inspired meta-heuristics are being proposed to address various open challenges. Since many heuristics are stochastic, they could yield different solutions to the same problem for different runs. Hence, there is a need for stringent in-depth statistical analysis of the performances of stochastic optimization algorithms. The proposed severity framework enables researchers and practitioners to define application-specific and meaningful performance evaluation metric that evaluates the magnitude of the performance improvement achieved, which is not only of statistical significance but also of practical relevance.

ABSTRACT. Managing population diversity in the optimization process in Evolutionary Algorithms can be a determinant factor for the quality of solutions found. Due to diverse problem characteristics, many techniques face difficulties and converge prematurely in local optima. The maintenance of diversity allows the algorithm to explore the search space and efficiently achieve better results. Parallel models are well-known techniques to maintain population diversity; however, design choices lead to different characteristics for the optimization process. For instance, the migration policy on the Island model can control how fast the algorithm converges. This work proposes and explores a fitness-based migration policy designed for the Biased Random-Key Genetic Algorithm (BRKGA). The proposal is compared with two traditional strategies and evaluates its performance in continuous search spaces. The results show that the proposal can improve the BRKGA optimization capability with suitable parameters.

ABSTRACT. Exploratory landscape analysis in single-objective black-box optimization relies on a comprehensive and large set of numerical features characterizing problem instances. Those foster problem understanding and serve as basis for constructing automated algorithm selection models choosing the best suited algorithm for a problem at hand based on the aforementioned features computed prior to optimization. This work specifically points to the sensitivity of a substantial proportion of these features to absolute objective values, i.e., we observe a lack of shift and scale invariance. We show that this unfortunately induces bias within automated algorithm selection models, an overfitting to specific benchmark problem sets used for training and thereby hinders generalization capabilities to unseen problems. We tackle these issues by presenting an appropriate objective normalization to be used prior to ELA feature computation and empirically illustrate the respective effectiveness focusing on the BBOB benchmark set.

ABSTRACT. A large number of airports across Europe are resource constrained. With long-term growth in air transportation expected to rise, more airports are expected to feel the imbalance between increased demand and limited resource capacity. This imbalance will lead to increasingly constrained scenarios, which require sophisticated solution methods to produce viable schedules. The use of ‘slots’ is one of the core mechanisms for managing access to constrained resources at airports. This paper presents a genetic algorithm-based hyper-heuristic approach to construct feasible solutions to the single airport slot allocation problem. To evaluate the proposed approach, we compare the solutions found by a number of previously developed and newly proposed constructive heuristics, over a range of real-world data sets. Our results show that the hyper-heuristic outperforms any individual constructive heuristic in all test instances, overcoming the drawbacks that a single heuristic faces when required to solve instances with different problem features.

ABSTRACT. The scheme followed to respond to the environmental change when a change is detected is the key that distinguishes different algorithms proposed to solve the dynamic multi-objective optimization problems. When it comes to the situation of detecting a change in the environment, one of the most important characteristics to take into consideration is the severity of the change. Measuring the severity of change can clarify significant information about the change which can be used to respond effectively to the change. In this paper, a new framework has problems because the dynamism dramatically decreases the optimizer's performance. When a change is detected, the severity of the change is estimated and an appropriate approach is followed to react based on the estimated severity. Moreover, the algorithm responds multiple times for the same change to accelerate the convergence process. To examine the performance of the proposed algorithm, three state-of-the-art strategies are compared using six different benchmarks. The experimental results demonstrate the effectiveness of the proposed approach where it outperforms other algorithms in most of the tested instances.

ABSTRACT. Designing cable-stayed bridges requires the determination of several design variables' values. Civil engineers usually perform this task by hand as an iteration of steps that stops when the engineer is happy with both the cost and maintaining the structural constraints of the solution. The problem's difficulty arises from the fact that changing a variable may affect other variables, meaning that they are not independent, suggesting that we are facing a deceptive landscape. In this work, we compare two approaches to a baseline solution: a Genetic Algorithm and a CMA-ES algorithm. There are two objectives when designing the bridges: minimising the cost and maintaining the structural constraints in acceptable values to be considered safe. These are conflicting objectives, meaning that decreasing the cost often results in a bridge that is not structurally safe. The results suggest that CMA-ES is a better option for finding good solutions in the search space, beating the baseline with the same amount of evaluations, while the Genetic Algorithm could not. In concrete, the CMA-ES approach is able to design bridges that are cheaper and structurally safe.

ABSTRACT. The invariance to any monotonically increasing transformation of the objective function is an essential property of the covariance matrix adaptation evolution strategy (CMA-ES). However, the surrogate-assisted CMA-ES often loses this invariance because the performance of the surrogate model is influenced by such transformation. In this paper, we propose a surrogate-assisted (1+1)-CMA-ES with the Gaussian process regression (GPR) possessing the robustness for such transformation. We introduce two utility functions based on the ranking and Lebesgue measure of candidate solutions, which are invariant to such transformation. We train GPR with the utility function values instead of the objective function values to estimate the ranking of the candidate solutions. Moreover, we propose switching Gaussian process CMA-ES (SGP-CMA-ES), which contains a mechanism selecting the suitable target variable of GPR from utility function values in addition to the objective function values to improve the performance on the objective functions that GPR can estimate easily. The experimental results show that SGP-CMA-ES is superior to existing methods on the objective function with several transformations, while maintaining the performance comparable to that of existing methods on the objective function without transformation.

ABSTRACT. With the increasing demand for high-quality internet services, deploying GPON/Fiber-to-the-Home networks is one of the biggest challenges that internet providers have to deal with due to the significant investments involved. Automated network design usage becomes more critical to aid with planning the network by minimising the costs of planning and deployment. The main objective is to tackle this problem of optimisation of networks that requires taking into account multiple factors such as the equipment placement and their configuration, the optimisation of the cable routes, the optimisation of the clients' allocation and other constraints involved in the minimisation problem. An AI-based solution is proposed to automate network design, which is often done manually. It is a difficult task requiring a significant amount of time to complete by hand. To alleviate this tiresome task, we proposed a Genetic Algorithm using a two-level representation to design networks automatically. To validate the approach, we compare the quality of the generated solutions with the manual design ones. The results show that our method can save costs and time in finding suitable and better solutions than existing ones, indicating its potential as a support design tool of solutions for GPON/Fiber-to-the-Home networks. In concrete, in the two scenarios where we validate our proposal, we can cut costs by 31% in one and by 52.2% in the other, showcasing the potential of the proposed approach.

ABSTRACT. Taking levodopa, a drug used to treat symptoms of Parkinson's disease, is often connected with severe side effects, known as Levodopa-induced dyskinesia (LID). It can fluctuate in severity throughout the day and thus is difficult to classify during a short period of a physician's visit. A low-power wearable classifier enabling long-term and continuous LID classification would thus significantly help with LID detection and dosage adjustment. This paper deals with a co-evolutionary design of energy-efficient hardware accelerators of LID classifiers that can be implemented in wearable devices. The accelerator consists of a feature extractor and a classifier co-evolved using cartesian genetic programming (CGP). We introduce and evaluate a fast and accurate energy consumption estimation method for the target architecture of considered classifiers. The proposed energy estimation method allows for a multi-objective design enabled by introducing energy constraints. With the introduction of variable data representation bit width, the proposed method achieves a good trade-off between accuracy (AUC) and energy consumption.

ABSTRACT. Bent functions are a type of Boolean functions with properties that make them useful in cryptography. In this paper we propose a new semantic mutation operator for design of bent Boolean functions via genetic programming. To assess the efficiency of the proposed operator, we compare it to several other commonly used non-semantic mutation operators. Our results show that semantic mutation makes the evolutionary process more efficient, and significantly decreases the number of fitness function evaluations required to find a bent function.

ABSTRACT. Because it’s impossible to generate an initial population of valid individuals. When one million random instances are made, the number of constraint violations increases quadratically in instance size. Possibly, the density of feasible schedules gets extremely low for larger instances, not only prohibiting the mere creation of an initial population, but also operations such as crossover and mutation.

ABSTRACT. We consider a flow shop scheduling problem where the first machine which is a single machine and the second machine is a batch processing machine. The jobs are transported from the first to the second machine by a robot. The first machine can process only one job at a time and the second machine can process a batch of jobs simultaneously. The capacity of the robot is equal to the capacity of the batch processing machine. When the robot arrives at the second machine, all jobs contained in the same sequence define the delivered batch. Several variations of this model have been used in industry and in theoretical studies. Practice has shown, however, that the time between machines should be taken into consideration given its importance in the industry of today. Since the problem is NP-hard, we have proposed two heuristics to solve large instances. In the experimentations part, the tests prove the performance of the developed methods.

ABSTRACT. . The assignment problem (AP) is very close to the asymmetric traveling salesman problem (ATSP). So close even, that some AP solutions are ATSP solutions as well, without further amendments. This is interesting because the AP is solvable in polynomial time, and ATSP is not1. This partial replication study shows that for 100x100 matrices, approximately 2.017% of AP solutions are also ATSP solutions, but the percentage depends on the numerical entries inside the matrix in a nonconclusive way, as the results in both studies differ on an initial segment. The community is invited to hypothesize and discuss these results, as well as directions of further study

ABSTRACT. Recently, symbolic music generation with deep learning techniques has witnessed steady improvements. Most works on this topic focus on MIDI representations, but less attention has been paid to symbolic music generation using guitar tablatures (tabs) which can be used to encode multiple instruments. Tabs include information on expressive techniques and fingerings for fretted string instruments in addition to rhythm and pitch. In this work, we use the DadaGP dataset for guitar tab music generation, a corpus of over 26k songs in GuitarPro and token formats. We introduce methods to condition a Transformer-XL deep learning model to generate guitar tabs (GTR-CTRL) based on desired instrumentation (inst-CTRL) and genre (genre-CTRL). Special control tokens are appended at the beginning of each song in the training corpus. We assess the performance of the model with and without conditioning. We propose instrument presence metrics to assess the inst-CTRL model's response to a given instrumentation prompt. We trained a BERT model for downstream genre classification and used it to assess the results obtained with the genre-CTRL model. Statistical analyses evidence significant differences between the conditioned and unconditioned models. Overall, results indicate that the GTR-CTRL methods provide more flexibility and control for guitar-focused symbolic music generation than an unconditioned model.

ABSTRACT. Type Design is a domain that multiple times has profited from the emergence of new tools and technologies. The transformation of type from physical to digital, the dissemination of font design software and the adoption of web typography make type design both better known and more accessible. This domain has received an even greater push with the increasing adoption of generative tools to create more diverse and experimental fonts. Nowadays, with the application of Machine Learning to various domains, typography has also been influenced by it. In this work, we produce a dataset by extracting letter skeletons from a collection of existing fonts. Then we trained a Variational Autoencoder and a Sketch Decoder to learn to create these skeletons, which can be use to generate new ones by exploring the latent space. This process also allows us to control the style of the resulting skeleton and interpolate between different letters. Finally, we developed new glyphs by filling the generated skeletons based on the original letters' stroke width and show some applications of the results.

ABSTRACT. Sonification of visual information is a relatively new research line that aims to create a new way to access and experience visual displays, especially for the visually impaired. When applied to artworks, sonification needs to translate the aesthetic experience as well. This is attempted via a handful studies in the literature, where most of the transformation and music generation is done manually, or only by using the low level visual features of artworks. In this paper, we present a sonification model that uses both low level and high level features such as color, edge information, saliency, object and scene detection to create a pleasant and descriptive sonification of artworks with the use of a fully automatic pipeline. The results of the model are tested via interviews done with experts in music theory and generative music models. We found a high agreement among experts for the evaluation of a small set of sonified paintings. Addition of high level features such as scene sounds played a big role in this. Among the challenges observed during the interviews was the need to add emotion and mood information as well as semantic information to the sonification in order to create more descriptive melodies and sounds. The complexity and ambiguity of the visual information generated the most disagreement among experts both in their interpretation of the paintings as well as their sonifications.

ABSTRACT. Genotype-to-phenotype mappings translate genotypic variations such as mutations into phenotypic changes. Neutrality is the observation that some mutations do not lead to phenotypic changes. Studying the search trajectories in genotypic and phenotypic spaces, especially through neutral mutations, helps us to better understand the progression of evolution and its algorithmic behaviour. In this study, we visualise the search trajectories of a genetic programming system as graph-based models, where nodes are genotypes/phenotypes and edges represent their mutational transitions. We also quantitatively measure the characteristics of phenotypes including their genotypic abundance (the requirement for neutrality) and Kolmogorov complexity. We connect these quantified metrics with search trajectory visualisations, and find that more complex phenotypes are under-represented by fewer genotypes and are harder for evolution to discover. Less complex phenotypes, on the other hand, are over-represented by genotypes, are easier to find, and frequently serve as stepping-stones for evolution.

ABSTRACT. Cooperative Coevolution is a way to solve complex optimization problems by dividing them in smaller, simpler sub-problems. Those sub-problems are then tackled concurrently by evolving one population of solutions—actually, components of a larger solution—for each of them. However, components cannot be evaluated in isolation: in the common case of two concurrently evolving populations, each solution of one population must be coupled with another solution of the other population (the collaborator) in order to compute the fitness of the pair. Previous studies have already shown that the way collaborators are chosen and, if more than one is chosen, the way the resulting fitness measures are aggregated, play a key role in determining the success of coevolution. In this paper we perform an experimental analysis aimed at shedding new light on the effects of collaborators selection and aggregation. We first propose a general scheme for cooperative coevolution of two populations that allows to (a) use different EAs and solution representations on the two sub-problems and to (b) set different collaborators selection and aggregation strategies. Second, we instantiate this general scheme in a few variants and apply it to four optimization problems with different degrees of separability: two toy problems and two real prediction problems tackled with different kinds of model (symbolic regression and neural networks). We analyze the outcomes in terms of (a) effectiveness and efficiency of the optimization and (b) complexity and generalization power of the solutions. We find that the degree to which selection and aggregation schemes differ strongly depends on the interaction between the components of the solution.

ABSTRACT. Genetic programming is an evolutionary approach known for its performance in program synthesis. However, it is not yet mature enough for a practical use in real-world software development, since usually many training cases are required to generate programs that generalize to unseen test cases. As in practice, the training cases have to be expensively hand-labeled by the user, we need an approach to check the program behavior with a lower number of training cases. Metamorphic testing needs no labeled input/output examples. Instead, the program is executed multiple times, first on a given (randomly generated) input, followed by related inputs to check whether certain user-defined relations between the observed outputs hold. In this work, we suggest MTGP, which combines metamorphic testing and genetic programming and study its performance and the generalizability of the generated programs. Further, we analyze how the generalizability depends on the number of given labeled training cases. We find that using metamorphic testing combined with labeled training cases leads to a higher generalization rate than the use of labeled training cases alone in almost all studied configurations. Consequently, we recommend researchers to use metamorphic testing in their systems if the labeling of the training data is expensive.

ABSTRACT. The automatic generation of dispatching rules (DRs) for various scheduling problems using genetic programming (GP) has become an increasingly researched topic in recent years. Creating DRs in this way relieves domain experts of the tedious task of manually designing new rules, but also often leads to the discovery of better rules than those already available. However, developing new DRs is a computationally intensive process that takes time to converge to good solutions. A common way to improve the convergence of evolutionary algorithms is to use a more sophisticated method to generate the initial population of individuals. In this paper, we propose a simple method for initialising individuals that uses probabilistic information from previously evolved DRs. The method extracts the information on how many times each node occurs at each level of the tree and in each context. This information is then used to introduce bias in the selection of the node to be selected at a particular position during the construction of the expression tree. The experiments show that with the proposed method it is possible to improve the convergence of GP when generating new DRs, so that GP can obtain high-quality DRs in a much shorter time.

ABSTRACT. In Evolutionary Robotics where both body and brain are malleable, it is common practice to evaluate individuals in isolated environments. With the objective of implementing a more naturally plausible system, we designed a single interactive ecosystem for robots to be evaluated in. In this ecosystem robots are physically present and can interact each other and we implemented decentralized rules for mate selection and reproduction. To study the effects of evaluating robots in an interactive ecosystem has on evolution, we compare the evolutionary process with a more traditional, oracle--based approach. In our analysis, we observe how the different approach has a substantial impact on the final behaviour and morphology of the robots, while maintaining decent fitness performance.

ABSTRACT. With the increasing number of cloud services in multi-cloud, it has been a challenging task to choose suitable cloud services in consideration of multiple conflicting objectives. Multi-objective location-aware service brokering in multi-cloud aims to find a set of trade-off solutions that minimize both the cost and latency. To achieve this goal, existing approaches either manually design brokering heuristics or automatically generate heuristics via Genetic Programming Hyper-Heuristics (GPHH) for each problem domain from scratch. However, manually designing heuristics takes long time and requires domain knowledge. Also, knowledge learnt from one problem domain can be helpful for solving another problem domain. To effectively and efficiently generate heuristics for any new problem domain, we propose three novel GPHH-based approaches with transfer learning to automatically generate a group of Pareto-optimal heuristics. Experimental evaluations on real-world datasets demonstrate that our proposed GPHH with transfer learning approaches can outperform existing approaches.

ABSTRACT. The present of Internet of Things (IoT) devices has become essential in everyday human life. Because IoT devices often have small processing capability and low power supply, two popular technologies, i.e. cloud servers and fog edges, are increasingly integrated with IoT for workflow execution, giving rise to the resource allocation and workflow scheduling problem in hybrid IoT environments, i.e. the IoTWS problem. To tackle this NP-hard IoTWS problem, a new Genetic Algorithm(GA) called IoTGA has been successfully developed in this paper. In comparison to state-of-the-art GA approaches from literature, IoTGA allows fast workflow execution and can explicitly reduce the time and energy consumption thanks to its use of a newly designed local search method. Experiments on benchmark IoTWS problems clearly indicate that IoTGA can significantly outperform several competing GA methods and are more useful in practice.

ABSTRACT. With the integration of efficient computation and communication technologies into sensory devices, the Internet of Things (IoT) applications have increased tremendously in recent decades. While these applications provide numerous benefits to our daily lives, they also pose a great potential risk in terms of security. One of the reasons for this is that devices in IoT-based networks are highly resource constrained and interconnected over lossy links that can be exposed by attackers. The Routing Protocol for Low-Power and Lossy Network (RPL) is the standard routing protocol for such lossy networks. Despite the efficient routing built by RPL, this protocol is susceptible to insider attacks. Therefore, researchers have been working on developing effective intrusion detection systems for RPL-based IoT. However, most of these studies consume excessive resources (e.g., energy, memory, communication, etc.) and do not consider the constrained characteristics of the network. Hence, they might not be suitable for some devices/networks. Therefore, in this study, we aim to develop an intrusion detection system (IDS) that is both effective and efficient in terms of the cost consumed by intrusion detection (ID) nodes. For this multiple-objective problem, we investigate the use of evolutionary computation-based algorithms and show the performance of evolved intrusion detection algorithms against various RPL-specific attacks.

ABSTRACT. Evolutionary Algorithms (EAs) are routinely applied to solve a large set of optimization problems. Traditionally, their performance in solving those problems is analyzed using the fitness quality and computing time, and the effect of evolutionary operators on both metrics is frequently analyzed to compare different versions of EAs. Nevertheless, scientists face nowadays the challenge of considering energy efficiency in addition to computational time, which requires studying the algorithms energy consumption. This paper discusses the interest in introducing power consumption as a new metric to analyze the performance of genetic programming (GP). Two well-studied benchmark problems are addressed on three different computing platforms, and two different methods to measuring the power consumption have been tested. Analyzing the population size, the results demonstrate its influence on the energy consumed: a non-linear relationship was found between the size and energy required to complete an experiment. This study shows that not only computing time or solution quality must be analyzed, but also the energy required to find a solution. Summarizing, this paper shows that when GP is applied, specific considerations on how to select parameter values must be taken into account if the goal is to obtain solutions while searching for energy efficiency. Although the study has been performed using GP, we foresee that it could be similarly extended to EAs.

ABSTRACT. Nowadays, Cybersecurity is a significant concern that requires a significant amount of resources from both security companies and academia. One of the main research areas in this field is Intrusion Detection Systems (IDSs), which are designed to monitor network traffic and detect suspicious patterns or attacks on any node in the network. MSNM (Multivariate Statistical Network Monitoring) is an advanced algorithm that can effectively detect different types of security threats in real network traffic data. However, semi-supervised MSNM relies heavily on a set of weights, the values of which are determined using a relatively simple optimization algorithm. This paper proposes using different Evolutionary Algorithm approaches to optimize the set of variables, thereby improving the performance of MSNM against various attacks, including port scanning and botnets. To this end, a dataset called UGR’16, which is specifically designed to test IDSs, was used, and the performance of a Particle Swarm Optimization approach was analyzed. The results are very promising and suggest that EAs are a powerful tool for enhancing the performance of this IDS.

ABSTRACT. We present an artistic audio steganography technique for hiding stipple images inside of uncompressed audio that we dub ``stipple tunes.'' Given an audio carrier and a stipple image to hide, the goal is to manipulate samples in the left and right audio channels to draw the stipple points; that is, the left and right channels are interpreted, respectively, as X and Y coordinates in the Cartesian plane. To accomplish this, we devise an objective function that pans the audio and restricts samples to the stipple, while minimizing error, which we solve using the Viterbi algorithm. Decoding the hidden image is trivial; we simply create a scatterplot of the audio samples. We provide code, examples, and an interactive viewer in Javascript at http://stippletunes.s3-website-us-east-1.amazonaws.com/

ABSTRACT. In 2022, we organized the first Workshop on Evolutionary Computation and Explainable AI (ECXAI). With no pretence at completeness, this paper briefly comments on its outcome, what has happened since then in the field, and our expectations for the near future.

ABSTRACT. We consider the use of evolutionary algorithms (EAs) in byzantine environments in which fitness information can be computed by malicious agents. The performance of panmictic EAs is analyzed in this context, measuring the influence of the rate of unreliability of the environment. It is shown that even for simple problems there is noticeable performance degradation, highlighting the need for appropriate mechanisms to cope with this issue.