Résultats de recherche pour « Bayesian Deep Learning »

Abstract Traffic flow distribution forecasting plays a crucial role in intelligent transportation systems for effective traffic management and congestion mitigation. In this paper, we propose a novel approach for traffic flow distribution forecasting using federated learning and Bayesian deep learning. The proposed method leverages the collaborative training capability of federated learning to learn from decentralized data sources while preserving data privacy and ownership. Bayesian deep learning techniques are integrated to estimate uncertainties and provide probabilistic predictions, enabling the capture of the inherent variability in traffic flow distributions. We conduct comprehensive experiments on real-world traffic flow datasets and compare our method with baseline approaches. The results demonstrate that our proposed method outperforms the baselines in terms of mean absolute error, root mean squared error, negative log-likelihood, and calibration error. The integration of federated learning and Bayesian deep learning enables accurate traffic flow distribution forecasts while providing reliable uncertainty estimates. This work contributes to the advancement of intelligent transportation systems and provides valuable insights for decision-making in traffic management and congestion reduction.

<p><b>Inverse problems are among the most challenging and widespread problems in science today. Inverse problems are commonly ill-conditioned, having potentially infinitely many solutions, and prior assumptions must be made in order to make the solution unique. Techniques for solving inverse problems are typically bespoke, requiring different tools and frameworks for each problem.</b></p> <p>In this thesis we use Bayes' rule as a generic solution to a range of inverse problems, coupled with the deep learning technique known as normalising flows to provide models for the prior. Bayes' provides great flexibility -- being easily applicable to a range of inverse problems -- as well as results competitive with, or even exceeding, state-of-the-art techniques in image reconstruction, MRI acceleration and radio astronomy.</p>

Deep reinforcement learning (DRL) algorithms interact with the environment and have achieved considerable success in several decision-making problems. However, DRL requires a significant number of data before it can achieve adequate performance. Moreover, it might have limited applicability when DRL agents are able to learn in a real-world environment. Therefore, some algorithms combine DRL agents with supervised learning and leverage previous additional knowledge. Some have integrated a deep Q-learning network with a behavioral cloning model that can exploit supervised learning as prior learning. The algorithm proposed in this study is also based on these methods and is intended to update the loss function of the existing technique into a Bayesian approach. The supervised loss function used in existing algorithms and the loss function based on the Bayesian method proposed in this study differ in terms of the utilization of prior knowledge. Using prior knowledge and not using prior knowledge, such as the cross entropy being symmetric. As a result of the various OpenAI Gym environments, such as Cart-Pole and MountainCar, the learning convergence performance was improved. In particular, the proposed method can be applied to achieve fairly stable learning during the early stage when learning in a sparse environment is uncertain.

Abstract Although deep neural networks have shown remarkable success in many applications, however, they are still quite difficult to deploy on devices with limited hardware resources, as they consume large amounts of memory and compute power. However, quantization methods can reduce memory consumption and increase computation efficiency, but usually lead to the loss of uncertainty estimation capability, which is an essential task in applications that require safety. In this paper, we introduce an innovative approach called Bayesian Asymmetric Quantized Deep Learning with Triangulation Topology Aggregation (BAQTTA), where we provide a novel optimization framework that integrates Bayesian uncertainty estimation with asymmetric quantization with a triangulation-level topology aggregation optimizer. To achieve uncertainty estimation in discrete weight space, we perform variational inference for discrete weight distributions, learn asymmetric bit width allocation for weights and activations, and propose an optimization strategy for Delaunay triangulation-structured gradient aggregation that utilizes the geometric properties in the weight space. Experimental evaluations on the CIFAR-10, CIFAR-100, ImageNet datasets demonstrate that BAQ-TTA reduces memory usage by up to 8 times with inference speedup of 3.2 times in comparison to full precision models with good uncertainty estimation performance. Our proposed triangulation topology aggregation optimizer reduces the total optimization time by 23% compared to normal optimizers in quantized learning conditions. Our approach focuses on using 4-bit weight and 8-bit activation. It achieves an accuracy of 94.2% on the CIFAR-10 and 76.8% Top-1 on ImageNet with improved negative log-likelihood performance of 12 to 18% over quantized learning.

Abstract With the rise of ride-sharing services, accurately predicting origin-destination demand has become a critical task for ensuring efficient allocation of resources and enhancing the overall user experience. Traditional demand prediction methods often rely on deterministic models that fail to capture the inherent uncertainty and variability in travel patterns. In this paper, we propose a novel approach for origin-destination demand prediction in ride-sharing using deep Bayesian learning techniques. By leveraging the power of deep neural networks and the flexibility of Bayesian modeling, our method not only provides accurate demand estimates but also quantifies the associated uncertainties. We demonstrate the effectiveness of our approach through extensive experiments on a real-world ride-sharing dataset, comparing it with several baseline methods commonly used in the field. The experimental results showcase the superior performance of our proposed method in accurately predicting the origin-destination demand in ride-sharing. Our approach achieves significantly lower mean absolute error (MAE) and root mean square error (RMSE) compared to the baseline methods, indicating its improved accuracy. Moreover, our method demonstrates a higher coverage probability, highlighting its better reliability and uncertainty estimation capabilities. The key contributions of our paper include the development of a deep Bayesian neural network for demand prediction, the introduction of probabilistic layers to capture uncertainties, and the utilization of dropout regularization for improved uncertainty estimation. Our proposed approach offers a robust and scalable solution for accurate origin-destination demand prediction in ride-sharing, providing valuable insights for decision-making and resource allocation.