This paper introduces a novel framework for replicating manual processes in digital twins. The proposed approach allows the modeling of manual process variability, a key aspect often overlooked in the literature. Our research highlights the potential of AI to tailor the digital twin to specific contexts influenced by human factors. The model's accuracy can be improved with each simulation synchronization cycle through supervised machine learning, which enhances the alignment of the virtual and physical processes. The proposed digital twin framework aims to avoid discrepancies from the physical counterpart, disregarding decisions based on fixed parameters that do not evolve over time and ensuring a higher degree of realism of the virtual replica. Ultimately, we aim to design the digital twin as a realistic and effective decision-making aid for all human participants engaged in the digital twin loop.

The ever-increasing complexity of both Deep Neural Networks (DNN) and hardware accelerators has made the co-optimization of these domains extremely complex. Previous works typically focus on optimizing DNNs given a fixed hardware configuration or optimizing a specific hardware architecture given a fixed DNN model. Recently, the importance of the joint exploration of the two spaces drew more and more attention. Our work targets the co-optimization of DNN and hardware configurations on edge GPU accelerators. We propose an evolutionary-based co-optimization strategy by considering three metrics: DNN accuracy, execution latency, and power consumption. By combining the two search spaces, a larger number of configurations can be explored in a short time interval. In addition, a better tradeoff between DNN accuracy and hardware efficiency can be obtained. Experimental results show that the co-optimization outperforms the optimization of DNN for fixed hardware configuration with up to 53% hardware efficiency gains with the same accuracy and inference time.

Convolutional Neural Networks (CNNs) are currently widely used in various fields, particularly for computer vision applications. Edge platforms have drawn tremendous attention from academia and industry due to their ability to improve execution time and preserve privacy. However, edge platforms struggle to satisfy CNNs’ needs due to their computation and energy constraints. Thus, it is challenging to find the most efficient CNN that respects accuracy, time, energy, and memory footprint constraints for a target edge platform. Furthermore, given the size of the design space of CNNs and hardware platforms, performance evaluation of CNNs entails several efforts. Consequently, designers need tools to quickly explore large design space and select the CNN that offers the best performance trade-off for a set of hardware platforms. This article proposes a Machine Learning (ML)–based modeling approach for CNN performances on edge GPU-based platforms for vision applications. We implement and compare five of the most successful ML algorithms for accurate and rapid CNN performance predictions on three different edge GPUs in image classification. Experimental results demonstrate the robustness and usefulness of our proposed methodology. For three of the five ML algorithms — XGBoost, Random Forest, and Ridge Polynomial regression — average errors of 11%, 6%, and 8% have been obtained for CNN inference execution time, power consumption, and memory usage, respectively.

Edge computing is increasingly used for Artificial Intelligence (AI) purposes to meet latency, privacy, and energy challenges. Convolutional Neural networks (CNN) are more frequently deployed on Edge devices for several applications. However, due to their constrained computing resources and energy budget, Edge devices struggle to meet CNN's latency requirements while maintaining good accuracy. It is, therefore, crucial to choose the CNN with the best accuracy and latency trade-off while respecting hardware constraints. This paper presents and compares five of the widely used Machine Learning (ML) based approaches to predict CNN's inference execution time on Edge GPUs. For these 5 methods, in addition to their prediction accuracy, we also explore the time needed for their training and their hyperparameters' tuning. Finally, we compare times to run the prediction models on different platforms. The use of these methods will highly facilitate design space exploration by quickly providing the best CNN on a target Edge GPU. Experimental results show that XGBoost provides an interesting average prediction error even for unexplored and unseen CNN architectures. Random Forest depicts comparable accuracy but needs more effort and time to be trained. The other 3 approaches (OLS, MLP, and SVR) are less accurate for CNN performance estimation.

Searching for an available parking space is a stressful and time-consuming task, which leads to increasing traffic and environmental pollution due to the emission of gases. To solve these issues, various solutions relying on information technologies (e.g., wireless networks, sensors, etc.) have been deployed over the last years to help drivers identify available parking spaces. Several recent works have also considered the use of historical data about parking availability and applied learning techniques (e.g., machine learning, deep learning) to estimate the occupancy rates in the near future. In this paper, we not only focus on training forecasting models for different types of parking lots to provide the best accuracy, but also consider the deployment of such a service in real conditions, to solve actual parking occupancy problems. It is therefore needed to continuously provide accurate information to the drivers but also to handle the frequent updates of parking occupancy data. The underlying challenges addressed in the present work so concern (1) the self-tuning of the forecasting model hyper-parameters according to the characteristics of the considered parking lots and (2) the need to maintain the performance of the forecasting model over time.