检索结果-内蒙古大学图书馆

8th CAA International Conference on Vehicular Control and Intelligence, CVCI 2024

作者： Sun, Jiawen Li, Wenyu Lin, Qiuting Duan, Feng Li, Zhenning Li, Shengbo Eben College of Artificial Intelligence Nankai University Tianjin China Nankai University Tianjin Key Laboratory of Ibci&ir Tianjin China University of Macau State Key Laboratory of IoTSC China School of Vehicle and Mobility College of Ai Tsinghua University Beijing China

ISBN: (纸本)9798331504892

Nowadays, human-vehicle cooperative driving is still an important solution for intelligent transportation in unstructured scenarios. However, the conflict between the driver and the automated system for multi-lane decision-making remains to be solved. When leaving the intended path, driver's expected trajectory cannot be identified by the automated system, re-sulting in safety hazards related to cooperation failure under incomplete decision information. This paper proposes a trajectory probability game control (TPGC) method to model the situation and intend to relieve the conflict. A Bayesian Nash equilibria strategy is derived using the proposed probabilistic dual iterative approximate dynamic programming (PADP). The equilibrium indicates a stable state for the human-vehicle system given the probability of driver's trajectory selection, where the uncertainty of the driver is reduced. A multi-lane simulation scenario is constructed to verify the effectiveness of the proposed control and solution method. The experimental results indicate that the designed controller can be applied to various trajectory selections and maintain an average lateral error within 0.17m on multi-lane roads. © 2024 IEEE.

关键词： approximate dynamic programming human-vehicle cooperative driving incomplete information game trajectory probability game control

来源：评论

学校读者我要写书评

暂无评论

dynamic Learning and Decision Making via Basis Weight Vectors

引用

OPERATIONS RESEARCH 2022年第3期70卷 1835-1853页

作者： Zhang, Hao Univ British Columbia Sauder Sch Business Vancouver BC V6T IZ2 Canada

This paper presents a new methodology to solve a general model of dynamic decision making with a continuous unknown parameter or state. The methodology centers on the "continuation-value functions" (mappings from the parameter space to the continuation-value space), created by feasible continuation policies. When the model primitives can be described through a family of basis functions (e.g., polynomials), a continuation-value function retains that property and can be represented by a basis weight vector. The set of efficient basis weight vectors can be constructed through backward induction, which leads to a significant reduction of problem complexity and enables an exact solution for small-sized problems. A set of approximation methods based on the new methodology is developed to tackle larger problems. The methodology is also extended to the multidimensional (multiparameter) setting, which features the problem of contextual multiarmed bandits with linear expected rewards. The approximation algorithm developed in this paper outperforms three benchmark algorithms (epsilon-greedy, Thompson sampling, and LinUCB) in learning situations with many actions and short horizons.

关键词： learning and doing dynamic pricing with learning linear contextual bandits approximate dynamic programming basis representation of functions

来源：评论

学校读者我要写书评

暂无评论

A Numerical Method for Hedging Bermudan Options under Model Uncertainty

引用

METHODOLOGY AND COMPUTING IN APPLIED PROBABILITY 2022年第2期24卷 893-916页

作者： Imai, Junichi Keio Univ Fac Sci & Technol Kohoku Ku 3-14-1 Hiyoshi Yokohama Kanagawa 2238522 Japan

Model uncertainty has recently been receiving more attention than risk. This study proposes an effective computational framework to derive optimal strategies for obtaining the upper and lower bounds of Bermudan-style options in the presence of model uncertainty. The optimal hedging strategy under model uncertainty can be formulated as a solution of a minimax problem. We employ approximate dynamic programming and propose an algorithm for effectively solving the minimax problem. This study considers a geometric Brownian motion and an exponential generalized hyperbolic Levy process as reference models. To take model uncertainty into consideration, we consider a set of equivalent probability measures via an Esscher or a class-preserving transform. Using numerical examples, we discuss the effects of model uncertainty on the size of tracking errors, the hedge portfolio, the possibility of early exercise and positions of options. In addition to investors' optimal strategies, the study examines Nature's optimal choice for equivalent probability measures. We find several notable phenomena that occur because of the existence of model uncertainty. We further examine the effects of different types of model uncertainty on option values and optimal hedging strategies.

关键词： Derivatives Hedging Model Uncertainty Financial Modeling approximate dynamic programming

来源：评论

学校读者我要写书评

暂无评论

Policy search for active fault diagnosis with partially observable state

引用

INTERNATIONAL JOURNAL OF ADAPTIVE CONTROL AND SIGNAL PROCESSING 2022年第9期36卷 2190-2216页

作者： Kral, Ladislav Puncochar, Ivo Univ West Bohemia Dept Cybernet Plzen Czech Republic

The article deals with a novel design of an active fault detector (AFD) for a nonlinear stochastic system with a partially observable state. The imperfect state information problem is converted to a perfect state information problem using a state estimator. Subsequently, the problem is decomposed into separate tasks of an optimal fault detector design and an approximate input generator design using a dynamic programming technique. While the former task is straightforward, the latter represents a nonlinear functional optimization problem. The input generator is approximated by a multi-layer perceptron neural network, and its unknown parameters are found using the policy search method. Effectiveness of the proposed AFD design is demonstrated numerically on a pendulum system and a heating/cooling system.

关键词： approximate dynamic programming fault detection neural networks reinforcement learning state estimation

来源：评论

学校读者我要写书评

暂无评论

approximate Information State for approximate Planning and Reinforcement Learning in Partially Observed Systems

引用

JOURNAL OF MACHINE LEARNING RESEARCH 2022年第1期23卷 1-83页

作者： Subramanian, Jayakumar Sinha, Amit Seraj, Raihan Mahajan, Aditya Adobe Syst India Private Ltd Digital Experience Cloud Media & Data Sci Res Lab Plot A5Noida Greater Noida ExpresswayBlock A Noida 201304 Uttar Pradesh India McGill Univ Dept Elect & Comp Engn 3480 Univ St Montreal PQ H3A 0E9 Canada

We propose a theoretical framework for approximate planning and learning in partially observed systems. Our framework is based on the fundamental notion of information state. We provide two definitions of information state i) a function of history which is sufficient to compute the expected reward and predict its next value;ii) a function of the history which can be recursively updated and is sufficient to compute the expected reward and predict the next observation. An information state always leads to a dynamic programming decomposition. Our key result is to show that if a function of the history (called approximate information state (AIS)) approximately satisfies the properties of the information state, then there is a corresponding approximate dynamic program. We show that the policy computed using this is approximately optimal with bounded loss of optimality. We show that several approximations in state, observation and action spaces in literature can be viewed as instances of AIS. In some of these cases, we obtain tighter bounds. A salient feature of AIS is that it can be learnt from data. We present AIS based multi-time scale policy gradient algorithms and detailed numerical experiments with low, moderate and high dimensional environments.

关键词： Partially observed reinforcement learning partially observable Markov decision processes approximate dynamic programming information state approximate information state

来源：评论

学校读者我要写书评

暂无评论

An Improved ADP Algorithm for Multistage Unit Commitment of Renewable Energy Integrated Power System 7

An Improved ADP Algorithm for Multistage Unit Commitment of ...

引用

7th IEEE Student Conference on Electric Machines and Systems, SCEMS 2024

作者： Zhu, Wenkai Lan, Jing Chen, Yu Electric Power Dispatching Control Center Guangdong Power Grid Corporation Guangzhou China School of Electric Power Engineering South China University of Technology Guangzhou China College of Science Minzu University of China Beijing China

ISBN: (纸本)9798350366945

With the continuous penetration of renewable energy sources (RESs), the unit commitment (UC), as a key operational problem in power system, is facing increasingly challenges due to the uncertainties. This paper proposes a decomposed optimization algorithm for the multistage UC with the uncertainties of RESs. The Wasserstein metric-based distribution ally robust optimization (DRO) is extensively applied to handle the uncertainties in UC, and the UC is decomposed into a series of nested subproblems by improved approximate dynamic programming (ADP). Compared with traditional ADP algorithm, the proposed algorithm improves Lagrange cut-based value function to adaptively tackle integer variables, while provides tractable and favorable robust commitment strategies. Moreover, a parallel technique is further used to accelerate the algorithm in both scenario and stage levels. Case studies on several test systems illustrate that the proposed algorithm is favorable in terms of computational accuracy and efficiency. © 2024 IEEE.

关键词： approximate dynamic programming distributionally robust Lagrange cut parallel computing unit commitment

来源：评论

学校读者我要写书评

暂无评论

Radar-Communications Waveform Co-Design Over-the-Air Using the WISCANet SDR Network

Radar-Communications Waveform Co-Design Over-the-Air Using t...

引用

IEEE Radar Conference (RadarConf)

作者： Doly, Shammi A. Chiriyath, Alex R. Herschfelt, Andrew Holtom, Jacob Ragi, Shankarachary Bliss, Daniel W. Arizona State Univ Sch Elect Comp & Energy Engn Tempe AZ 85287 USA Arizona State Univ Ctr Wireless Informat Syst & Computat Architectur Tempe AZ 85281 USA South Dakota Sch Mines & Technol Dept Elect Engn Rapid City SD 57701 USA

ISBN: (纸本)9781665436694

Many waveform co-design studies demonstrate theoretical performance enhancements but rarely provide a clear path toward implementing a tractable solution. To address this limitation, we developed a waveform co-design technique to maximize the joint radar-communications network's joint performance and a computationally tractable method for optimizing it. This waveform co-design technique is based on the theory of partially-observable Markov decision processes (POMDPs), which we solve using an approximate dynamic programming approach called nominal belief-state optimization (NBO). The POMDP framework's natural look-ahead feature allows us to trade between the short-term and long-term performance of both radar and communications tasks, which allows it to adapt to changes in system requirements and environmental conditions. Using the WISCANet over-the-air experimental radio testbed, we implement a simple joint radar-communications system and demonstrate this waveform design and optimization technique in a hardware-in-the-loop over-the-air demonstration. We further extend the problem by proposing a real-time waveform optimization solution using a Kalman Filter in a dynamic environment.

关键词： Spectrum sharing cognitive radar waveform co-design RF convergence spectral coexistence target tracking approximate dynamic programming

来源：评论

学校读者我要写书评

暂无评论

Adaptive robust control of the continuous-time two-input systems with unknown disturbance based on Q-function 12

Adaptive robust control of the continuous-time two-input sys...

引用

IEEE 12th Data Driven Control and Learning Systems Conference (DDCLS)

作者： Lv, Yongfeng Cui, Zhengyu Wang, Minlin Taiyuan Univ Technol Coll Elect & Power Engn Taiyuan 030024 Peoples R China Lanzhou Univ Technol Coll Elect & Informat Engn Lanzhou 730050 Peoples R China

ISBN: (纸本)9798350321050

Considering overshoot and chatter of the multi-input system with unknown interference, this paper studies the adaptive robust optimal controls of continuous-time two-input systems with an approximate dynamic programming (ADP) based Q-function scheme. A complex Hamilton-Jacobi-Issacs (HJI) equation is obtained with the two-input system and the zero-game theory, where a value function is constructed. Solving the HJI equation is a challenging task. Thus, an ADP-based Q-function with a neural network is constructed to learn the saddle point of the HJI equation. Simultaneously, an integral reinforcement signal of the critic networks is introduced such that the system drift and input dynamics in the HJI equation are relaxed when studying the saddle-point intractable solution. Then, the adaptive robust optimal actor and worst disturbance are approximated with another three networks. Finally, an F-16 aircraft plant is used to verify the proposed ADP-based Q-function.

关键词： Robust control adaptive control approximate dynamic programming multi-input system

来源：评论

学校读者我要写书评

暂无评论

Interpretable Optimal Stopping

引用

MANAGEMENT SCIENCE 2022年第3期68卷 1616-1638页

作者： Ciocan, Dragos Florin Misic, Velibor V. European Inst Business Adm INSEAD F-77305 Fontainebleau France Univ Calif Los Angeles Anderson Sch Management Los Angeles CA 90095 USA

Optimal stopping is the problem of deciding when to stop a stochastic system to obtain the greatest reward, arising in numerous application areas such as finance, healthcare, and marketing. State-of-the-art methods for high-dimensional optimal stopping involve approximating the value function or the continuation value and then using that approximation within a greedy policy. Although such policies can perform very well, they are generally not guaranteed to be interpretable;that is, a decision maker may not be able to easily see the link between the current system state and the policy's action. In this paper, we propose a new approach to optimal stopping wherein the policy is represented as a binary tree, in the spirit of naturally interpretable tree models commonly used in machine learning. We show that the class of tree policies is rich enough to approximate the optimal policy. We formulate the problem of learning such policies from observed trajectories of the stochastic system as a sample average approximation (SAA) problem. We prove that the SAA problem converges under mild conditions as the sample size increases but that, computationally, even immediate simplifications of the SAA problem are theoretically intractable. We thus propose a tractable heuristic for approximately solving the SAA problem by greedily constructing the tree from the top down. We demonstrate the value of our approach by applying it to the canonical problem of option pricing, using both synthetic instances and instances using real Standard & Pool's 500 Index data. Our method obtains policies that (1) outperform state-of-the-art noninterpretable methods, based on simulation regression and martingale duality, and (2) possess a remarkably simple and intuitive structure.

关键词： optimal stopping approximate dynamic programming interpretability decision trees option pricing

来源：评论

学校读者我要写书评

暂无评论

Opaque selling of multiple substitutable products with finite inventories

引用

NAVAL RESEARCH LOGISTICS 2022年第4期69卷 529-549页

作者： Liu, Qian Xiao, Yongbo Zhang, Dan Hong Kong Univ Sci & Technol Dept Ind Engn & Decis Analyt Hong Kong Peoples R China Tsinghua Univ Res Ctr Contemporary Management Sch Econ & Management Beijing 100084 Peoples R China Univ Colorado Leeds Sch Business Boulder CO 80309 USA

Opaque selling, in which a seller offers opaque goods (OGs), in addition to physical goods, has been shown to be an effective strategy to segment a market and improve the seller's profit. This article studies opaque selling with stochastic demand and fixed initial inventories of multiple products, where the seller dynamically controls the product offers and determines the product assignment to fulfill the demand for OGs over time. The problem is formulated as a stochastic dynamic program. Due to the curse of dimensionality, we study the fluid control problem that gives a time-based fluid policy and a stationary probabilistic fulfillment strategy. We show that the fluid policy is asymptotically optimal when the arrival rates and initial inventory level are scaled up linearly. Furthermore, we propose a decomposition heuristic based on the corresponding fluid solution. The decomposition heuristic is shown to provide a tighter upper bound than the fluid control problem. Numerical study on a set of test instances illustrates the performance and efficacy of opaque selling.

关键词： approximate dynamic programming asymptotic optimality customer choice behavior opaque selling

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：