Search

Large language models (LLMs) have taken the scientific world by storm, changing the landscape of natural language processing and human-computer interaction. These powerful tools can answer complex questions and, surprisingly, perform challenging creative tasks (e.g., generate code and applications to solve problems, write stories, pieces of music, etc.). In this paper, we present a collaborative game design framework that combines interactive evolution and large language models to simulate the typical human design process. We use the former to exploit users' feedback for selecting the most promising ideas and large language models for a very complex creative task - the recombination and variation of ideas. In our framework, the process starts with a brief and a set of candidate designs, either generated using a language model or proposed by the users. Next, users collaborate on the design process by providing feedback to an interactive genetic algorithm that selects, recombines, and mutates the most promising designs. We evaluated our framework on three game design tasks with human designers who collaborated remotely., (Submitted)

Minimalist game design was introduced a decade ago as a general design principle with a list of key properties for minimalist games: basic controls, simple but aesthetically pleasing visuals, interesting player choices with vast possibility spaces, and sounds that resonate with the design. In this paper, we present an experiment we did to explore minimalism in games using a bottom-up approach. We invited a small group of professional game designers and a larger group of game design students to participate in a seminal experiment on minimalism in game design. We started from the most basic game elements: one pixel and one key which provide the least amount of information we can display and reasonably the most elementary action players can perform. We designed a game that starts with a black pixel and asks players to press a key when the pixel turns white. This minimal game, almost a Skinner box, captures the essential elements of the mechanics of games like "The Impossible Game," which asks players to do nothing more than press a key at the right moment. We presented this game concept to the professional game designers and challenged them to create other games with the least amount of player interaction and displayed information. We did not specify any constraints (as usually done in other contexts) and left them free to express their view of minimalistic game design. We repeated the experiment with 100+ students attending a master-level course on video game design and development at our institution. We then analyzed the creations of the two groups, discussing the idea of minimalistic design that emerges from the submitted game concepts.

Deep Learning has established in the latest years as a successful approach to address a great variety of tasks. Healthcare is one of the most promising field of application for Deep Learning approaches since it would allow to help clinicians to analyze patient data and perform diagnoses. However, despite the vast amount of data collected every year in hospitals and other clinical institutes, privacy regulations on sensitive data - such as those related to health - pose a serious challenge to the application of these methods. In this work, we focus on strategies to cope with privacy issues when a consortium of healthcare institutions needs to train machine learning models for identifying a particular disease, comparing the performances of two recent distributed learning approaches - Federated Learning and Split Learning - on the task of Automated Chest X-Ray Diagnosis. In particular, in our analysis we investigated the impact of different data distributions in client data and the possible policies on the frequency of data exchange between the institutions., 8 pages, 1 figure

This chapter discusses the challenges and opportunities of procedural content generation (PCG) in games. It starts with defining three grand goals of PCG, namely multi-level multicontent PCG, PCG-based game design and generating complete games. The way these goals are defined, they are not feasible with current technology. Therefore we identify nine challenges for PCG research. Work towards meeting these challenges is likely to take us closer to realising the three grand goals. In order to help researchers get started, we also identify five actionable steps, which PCG researchers could get started working on immediately.

This manual describes the competition software for the Simulated Car Racing Championship, an international competition held at major conferences in the field of Evolutionary Computation and in the field of Computational Intelligence and Games. It provides an overview of the architecture, the instructions to install the software and to run the simple drivers provided in the package, the description of the sensors and the actuators.

An automated technique has recently been proposed to transfer learning in the hierarchical Bayesian optimization algorithm (hBOA) based on distance-based statistics. The technique enables practitioners to improve hBOA efficiency by collecting statistics from probabilistic models obtained in previous hBOA runs and using the obtained statistics to bias future hBOA runs on similar problems. The purpose of this paper is threefold: (1) test the technique on several classes of NP-complete problems, including MAXSAT, spin glasses and minimum vertex cover; (2) demonstrate that the technique is effective even when previous runs were done on problems of different size; (3) provide empirical evidence that combining transfer learning with other efficiency enhancement techniques can often yield nearly multiplicative speedups., Accepted at Parallel Problem Solving from Nature (PPSN XII), 10 pages. arXiv admin note: substantial text overlap with arXiv:1201.2241

Learning Classifier Systems (LCS) are population-based reinforcement learners used in a wide variety of applications. This paper presents a LCS where each traditional rule is represented by a spiking neural network, a type of network with dynamic internal state. We employ a constructivist model of growth of both neurons and dendrites that realise flexible learning by evolving structures of sufficient complexity to solve a well-known problem involving continuous, real-valued inputs. Additionally, we extend the system to enable temporal state decomposition. By allowing our LCS to chain together sequences of heterogeneous actions into macro-actions, it is shown to perform optimally in a problem where traditional methods can fail to find a solution in a reasonable amount of time. Our final system is tested on a simulated robotics platform., 20 pages

We address the problem of automatically designing maps for first-person shooter (FPS) games. An efficient solution to this procedural content generation (PCG) problem could allow the design of FPS games of lower development cost with near-infinite replay value and capability to adapt to the skills and preferences of individual players. We propose a search-based solution, where maps are evolved to optimize a fitness function that is based on the players’ average fighting time. For that purpose, four different map representations are tested and compared. Results obtained showcase the clear advantage of some representations in generating interesting FPS maps and demonstrate the promise of the approach followed for automatic level design in that game genre., peer-reviewed

Von Neumann, the inventor of the modern computer, realized that if life is a physical process, it should be possible to implement life in other media than biochemistry. In the 1950s, he was one of the first to propose the possibility of implementing genuine living processes in computers and robots. This perspective, while still controversial, is rapidly gaining momentum in many science and engineering communities. Below, we summarize our recent activities to create artificial life from scratch in physicochemical systems. We also outline the nature of the grand science and engineering challenges faced as we seek to realize Von Neumann's vision: Integration of information processing and material production from the nano- to the macroscale in technical systems.

International audience; After two papers comparing ATNoSFERES with XCSM, a Learning Classifier System with internal states, this paper is devoted to a comparison between ATNoSFERES and ACS (an Anticipatory Learning Classifier System). As previously, we focus on the way perceptual aliazing problems encountered in non-Markov environments are solved with both kinds of systems. We shortly present ATNoSFERES, a framework based on an indirect encoding Genetic Algorithm which builds finite-state automata controllers, and we compare it with ACS through two benchmark experiments. The comparison shows that the difference in performance between both system depends on the environment. This raises a discussion of the adequacy of both adaptive mechanisms to particular subclasses of non-Markov problems. Furthermore, since ACS converges much faster than ATNoSFERES, we discuss the need to introduce learning capabilities in our model. As a conclusion, we advocate for the need of more experimental comparisons between different systems in the Learning Classifier System community.

International audience; In this paper we present ATNoSFERES, a new framework based on an indirect encoding Genetic Algorithm which builds finite-state automata controllers able to deal with perceptual aliazing. In the context of our ongoing line of research, we compare it with XCSM, a memory-based extension of the most studied Learning Classifier System, XCS, through two benchmark experiments. We focus in particular on internal state generalization, and add special purpose features to ATNoSFERES to fulfill that comparison. We then discuss the role played by internal state generalization in the experiments studied.