Résultat de la recherche
Investigating Hyperparameter Optimization and Transferability for ES-HyperNEAT: A TPE Approach
Neuroevolution of Augmenting Topologies (NEAT) and its advanced version, Evolvable-Substrate HyperNEAT (ES-HyperNEAT), have shown great potential in developing neural networks. However, their effectiveness heavily depends on the selection of hyperparameters. This study investigates the optimization of ES-HyperNEAT hyperparameters using the Tree-structured Parzen Estimator (TPE) on the MNIST classification task, exploring a search space of over 3 billion potential combinations. TPE effectively navigates this vast space, significantly outperforming random search in terms of mean, median, and best accuracy. During the validation process, the best hyperparameter configuration found by TPE achieves an accuracy of 29.00% on MNIST, surpassing previous studies while using a smaller population size and fewer generations. The transferability of the optimized hyperparameters is explored in logic operations and Fashion-MNIST tasks, revealing successful transfer to the more complex Fashion-MNIST problem but limited to simpler logic operations. This study emphasizes a method to unlock the full potential of neuroevolutionary algorithms and provides insights into the hyperparameters’ transferability across tasks of varying complexity.
Assessment of the Efficiency of Public Hospitals in Romania
Considering the high pressure on the healthcare limited resources, mainly on hospitals, determined by the population ageing, and the increased incidence of chronic and infectious diseases, it is essential to both decrease expenditures and provide good quality healthcare. In this paper we focus on the efficiency of Romanian public hospitals. Our research goals are to identify and examine the inefficient public hospitals in Romania; to determine sources of inefficiency in Romanian public hospitals; to describe a potential reduction in all inputs on average to rationalize hospital resources; and recommend that hospital management be improved. We propose an approach that contains preliminary data analyses to obtain homogeneous distributions, then we use Data Envelopment Analysis to estimate the technical efficiency scores for the hospitals in the sample. The results showed that more than half of the examined small hospitals were technically inefficient and that they could have produced a larger number of discharges and consequently an increased number of inpatient days. Possible reductions in inputs were also indicated. These results suggest ways of improving hospital management and restructuring and reorganizing decisions that can be implemented in the hospital network.
Design principles of a central metadata repository as a key element of an integrated health information system
The Swiss Health System is a complex system with different groups of actors for which data is collected and analyzed using various methods, leading to a large heterogeneity and dispersion of available data. The specificities of a Swiss Health System favor a hybrid infrastructure to manage the heterogeneity and dispersion of Swiss health-related data. This policy brief shows the importance of a metadata management infrastructure to identify and describe health data resources and highlights several essential key elements for the design of a metadata repository and also raises important practical questions.
ASBAR: an Animal Skeleton-Based Action Recognition framework. Recognizing great ape behaviors in the wild using pose estimation with domain adaptation
To date, the investigation and classification of animal behaviors have mostly relied on direct human observations or video recordings with posthoc analysis, which can be labor-intensive, time-consuming, and prone to human bias. Recent advances in machine learning for computer vision tasks, such as pose estimation and action recognition, thus have the potential to significantly improve and deepen our understanding of animal behavior. However, despite the increased availability of open-source toolboxes and large-scale datasets for animal pose estimation, their practical relevance for behavior recognition remains under-explored. In this paper, we propose an innovative framework, To date, the investigation and classification of animal behaviors have mostly relied on direct human observations or video recordings with posthoc analysis, which can be labor-intensive, time-consuming, and prone to human bias. Recent advances in machine learning for computer vision tasks, such as pose estimation and action recognition, thus have the potential to significantly improve and deepen our understanding of animal behavior. However, despite the increased availability of open-source toolboxes and large-scale datasets for animal pose estimation, their practical relevance for behavior recognition remains under-explored. In this paper, we propose an innovative framework, ASBAR, for Animal Skeleton-Based Action Recognition, which fully integrates animal pose estimation and behavior recognition. We demonstrate the use of this framework in a particularly challenging task: the classification of great ape natural behaviors in the wild. First, we built a robust pose estimator model leveraging OpenMonkeyChallenge, one of the largest available open-source primate pose datasets, through a benchmark analysis on several CNN models from DeepLabCut, integrated into our framework. Second, we extracted the great ape’s skeletal motion from the PanAf dataset, a large collection of in-the-wild videos of gorillas and chimpanzees annotated for natural behaviors, which we used to train and evaluate PoseConv3D from MMaction2, a second deep learning model fully integrated into our framework. We hereby classify behaviors into nine distinct categories and achieve a Top 1 accuracy of 74.98%, comparable to previous studies using video-based methods, while reducing the model’s input size by a factor of around 20. Additionally, we provide an open-source terminal-based GUI that integrates our full pipeline and release a set of 5,440 keypoint annotations to facilitate the replication of our results on other species and/or behaviors. All models, code, and data can be accessed at: https://github.com/MitchFuchs/asbar.
Tool Wear Monitoring Using Multi-sensor Time Series and Machine Learning
In the milling process of micro-machining, the optimization process is one of the keys to reduce production cost. By monitoring the tool wear and detecting when it is no longer acceptable, the machining process can be adjusted more accurately. This research explores four approaches using different machine learning models to predict machining tool wear during the milling process. The study is based on a dataset created with a face milling operation on stainless steel (AISI 303) round material. The machining is divided into a number of stairs and is performed with a 3mm tungsten carbide. Three different types of sensors are used to measure the wearing process, with acoustic emission, accelerometers and axis currents. The better approach achieved a f1-score of 73% on five classes with a Extra Trees Classifier.
Swiss health metadata repository : implementation recommendations - requirements, recommended architecture and prototype implementation
In the context of a decentralized approach for health data resources, a central metadata repository becomes essential for the identification/description of the necessary resources. A Swiss Health Metadata Repository is expected to enable multiple advantages, namely: to provide a single entry point for searching/retrieving health-related resources; to help identify possible semantic linking between data sources; to provide a consistent data catalog ensured by the use of standardized vocabularies/ontologies; to enable potential exchange of experience and know-how between health data-related projects in Switzerland; and to increase the capacity of research groups to share/access/analyse health related data. This policy brief focuses on the implementation aspects of a proof-of-concept proto-type for Swiss Health Metadata Repository.