In recent years, reinforcement learning and its multi-agent analogue have achieved great success in solving various complex control problems. However, multi-agent reinforcement learning remains challenging both in its...
详细信息
ISBN:
(纸本)9798350323658
In recent years, reinforcement learning and its multi-agent analogue have achieved great success in solving various complex control problems. However, multi-agent reinforcement learning remains challenging both in its theoretical analysis and empirical design of algorithms, especially for large swarms of embodied robotic agents where a definitive toolchain remains part of active research. We use emerging state-of-the-art mean-field control techniques in order to convert many-agent swarm control into more classical single-agent control of distributions. This allows profiting from advances in single-agent reinforcement learning at the cost of assuming weak interaction between agents. However, the mean-field model is violated by the nature of real systems with embodied, physically colliding agents. Thus, we combine collision avoidance and learning of mean-field control into a unified framework for tractably designing intelligent robotic swarm behavior. On the theoretical side, we provide novel approximation guarantees for general mean-field control both in continuous spaces and with collision avoidance. On the practical side, we show that our approach outperforms multi-agent reinforcement learning and allows for decentralized open-loop application while avoiding collisions, both in simulation and real UAV swarms. Overall, we propose a framework for the design of swarm behavior that is both mathematically well-founded and practically useful, enabling the solution of otherwise intractable swarm problems.
This paper describes a learning approach based on training convolutional neural networks (CNN) for a traffic sign classification system. In addition, it presents the preliminary classification results of applying this...
详细信息
ISBN:
(纸本)9781538644492
This paper describes a learning approach based on training convolutional neural networks (CNN) for a traffic sign classification system. In addition, it presents the preliminary classification results of applying this CNN to learn features and classify RGB-D images task. To determine the appropriate architecture, we explore the transfer learning technique called "fine tuning technique", of reusing layers trained on the ImageNet dataset in order to provide a solution for a four -class classification task of a new set of data.
Traffic signal control is one of the most effective methods of traffic management in urban areas. In recent years, traffic control methods based on deep reinforcement learning (DRL) have gained attention due to their ...
详细信息
ISBN:
(纸本)9781728191423
Traffic signal control is one of the most effective methods of traffic management in urban areas. In recent years, traffic control methods based on deep reinforcement learning (DRL) have gained attention due to their ability to exploit real-time traffic data, which is often poorly used by the traditional hand-crafted methods. While most recent DRL-based methods have focused on maximizing the throughput or minimizing the average travel time of the vehicles, the fairness of the traffic signal controllers has often been neglected. This is particularly important as neglecting fairness can lead to situations where some vehicles experience extreme waiting times, or where the throughput of a particular traffic flow is highly impacted by the fluctuations of another conflicting flow at the intersection. In order to address these issues, we introduce two notions of fairness: delay-based and throughput-based fairness, which correspond to the two issues mentioned above. Furthermore, we propose two DRL-based traffic signal control methods for implementing these fairness notions, that can achieve a high throughput as well. We evaluate the performance of our proposed methods using three traffic arrival distributions, and find that our methods outperform the baselines in the tested scenarios.
Modern distributed cyber-physical systems encounter a large variety of anomalies and in many cases, they are vulnerable to catastrophic fault propagation scenarios due to strong connectivity among the sub-systems. In ...
详细信息
ISBN:
(纸本)9781509028733
Modern distributed cyber-physical systems encounter a large variety of anomalies and in many cases, they are vulnerable to catastrophic fault propagation scenarios due to strong connectivity among the sub-systems. In this regard, root-cause analysis becomes highly intractable due to complex fault propagation mechanisms in combination with diverse operating modes. This paper presents a new data-driven framework for root-cause analysis for addressing such issues. The framework is based on a spatiotemporal feature extraction scheme for distributed cyber-physical systems built on the concept of symbolic dynamics for discovering and representing causal interactions among subsystems of a complex system. We present two approaches for root-cause analysis, namely the sequential state switching (S-3, based on free energy concept of a Restricted Boltzmann Machine, RBM) and artificial anomaly association (A(3), a multi-class classification framework using deep neural networks, DNN). Synthetic data from cases with failed pattern(s) and anomalous node are simulated to validate the proposed approaches, then compared with the performance of vector autoregressive (VAR) model-based root-cause analysis. Real dataset based on Tennessee Eastman process (TEP) is also used for validation. The results show that: (1) S-3 and A(3) approaches can obtain high accuracy in root-cause analysis and successfully handle multiple nominal operation modes, and (2) the proposed tool-chain is shown to be scalable while maintaining high accuracy.
A batch learning method for competitive associative net called CAN2 is presented and applied to time series prediction of the CATS benchmark (for Competition on Artificial Time Series). Although we have presented onli...
详细信息
ISBN:
(纸本)0780383591
A batch learning method for competitive associative net called CAN2 is presented and applied to time series prediction of the CATS benchmark (for Competition on Artificial Time Series). Although we have presented online learning methods for the CAN2 so far, which are basically for infinite number of training data. Provided that only a finite number of training data are given, however, the batch learning scheme seems more suitable. We here present a batch learning method to efficiently learn a finite number of data. We finally apply the present method to the time series prediction of the CATS benchmark.
Dynamic data-driven applications aim to reconcile different sources of information for systems under scrutiny. Such problems ubiquitously arise in geosciences, for applications like numerical weather prediction, clima...
详细信息
ISBN:
(纸本)9781467366472
Dynamic data-driven applications aim to reconcile different sources of information for systems under scrutiny. Such problems ubiquitously arise in geosciences, for applications like numerical weather prediction, climate change and green energy harvesting. One of the main challenges in solving data-driven applications come from the associated large computational cost. This article presents an adaptive computational framework for fusing numerical model predictions with real observations, in order to generate discrete initial conditions which are optimal in a certain sense. The proposed framework incorporates four-dimensional variational data assimilation, observation impact via sensitivity analysis and adaptive measurement strategies.
Distribution grids are currently challenged by frequent voltage excursions induced by intermittent solar generation. Smart inverters have been advocated as a fast-responding means to regulate voltage and minimize ohmi...
详细信息
ISBN:
(纸本)9781728112954
Distribution grids are currently challenged by frequent voltage excursions induced by intermittent solar generation. Smart inverters have been advocated as a fast-responding means to regulate voltage and minimize ohmic losses. Since optimal inverter coordination may be computationally challenging and preset local control rules are subpar, the approach of customized control rules designed in a quasi-static fashion features as a golden middle. Departing from affine control rules, this work puts forth non-linear inverter control policies. Drawing analogies to multi-task learning, reactive control is posed as a kernel-based regression task. Leveraging a linearized grid model and given anticipated data scenarios, inverter rules are jointly designed at the feeder level to minimize a convex combination of voltage deviations and ohmic losses via a linearly-constrained quadratic program. Numerical tests using real-world data on a benchmark feeder demonstrate that nonlinear control rules driven also by a few non-local readings can attain near-optimal performance.
With increasing number of open-source deep learning (DL) software tools made available, benchmarking DL software frameworks and systems is in high demand. This paper presents design considerations, metrics and challen...
详细信息
ISBN:
(纸本)9781538668719
With increasing number of open-source deep learning (DL) software tools made available, benchmarking DL software frameworks and systems is in high demand. This paper presents design considerations, metrics and challenges towards developing an effective benchmark for DL software frameworks and illustrate our observations through a comparative study of three popular DL frameworks: TensorFlow, Caffe, and Torch. First, we show that these deep learning frameworks are optimized with their default configurations settings. However, the default configuration optimized on one specific dataset may not work effectively for other datasets with respect to runtime performance and learning accuracy. Second, the default configuration optimized on a dataset by one DL framework does not work well for another DL framework on the same dataset. Third, experiments show that different DL frameworks exhibit different levels of robustness against adversarial examples. Through this study, we envision that unlike traditional performance-driven benchmarks, benchmarking deep learning software frameworks should take into account of both runtime and accuracy and their latent interaction with hyper-parameters and data-dependent configurations of DL frameworks.
The manual dispensing of drugs by pharmacists is time-consuming, labor-intensive, and has a low accuracy rate. In order to address this issue, this paper proposes an automated drug dispensing method based on improved ...
详细信息
We consider discrete-time switching systems composed of a finite family of affine sub-dynamics. First, we recall existing results and present further analysis on the stability problem, the existence and characterizati...
详细信息
ISBN:
(纸本)9781665436595
We consider discrete-time switching systems composed of a finite family of affine sub-dynamics. First, we recall existing results and present further analysis on the stability problem, the existence and characterization of compact attractors, and the relations these problems have with the joint spectral radius of the set of matrices composing the linear part of the subsystems. Second, we tackle the problem of providing probabilistic certificates of stability along with the existence of forward invariant sets, assuming no knowledge on the system data but only observing a finite number of sampled trajectories. Some numerical examples illustrate the advantages and limits of the proposed conditions.
暂无评论