32,858 Hits in 9.0 sec

Image Classification by Reinforcement Learning with Two-State Q-Learning [article]

Abdul Mueed Hafiz
2020 arXiv   pre-print
In this paper, a simple and efficient Hybrid Classifier is presented which is based on deep learning and reinforcement learning.  ...  Here, Q-Learning has been used with two states and 'two or three' actions.  ...  For reinforcement learning, Q-Learning with random policy is used. Two states (n = 2) and 'two or three' actions (a = 2) or (a = 3) are used.  ... 
arXiv:2007.01298v3 fatcat:xvsd7voy6jgrhjaqqi4tlmpbla

An Empirical Analysis of Action Map in Learning Classifier Systems

2018 SICE Journal of Control Measurement and System Integration  
An action map is one of the most fundamental options in designing a learning classifier system (LCS), which defines how LCSs cover a state action space in a problem.  ...  It still remains unclear which action map can be adequate to solve which type of problem effectively, resulting in a lack of basic design methodology of LCS in terms of the action map.  ...  For instance, the accuracy based fitness of XCS is originally defined with the complete action map [6] ; a supervised learning technique of UCS is designed with the best action map.  ... 
doi:10.9746/jcmsi.11.239 fatcat:y4hnbkw34jaazlowfa22oltjta

Accuracy-Based Learning Classifier Systems: Models, Analysis and Applications to Classification Tasks

Ester Bernadó-Mansilla, Josep M. Garrell-Guiu
2003 Evolutionary Computation  
Departing from XCS, we analyze the evolution of a complete action map as a knowledge representation. We propose an alternative, UCS, which evolves a best action map more efficiently.  ...  This paper investigates two models of accuracy-based learning classifier systems on different types of classification problems.  ...  Our study focused on two models of accuracy-based LCS: XCS and UCS. XCS bases fitness on accuracy computed from a reinforcement learning scheme, and in doing so it evolves a complete action map.  ... 
doi:10.1162/106365603322365289 pmid:14558911 fatcat:25wu53rxhbe27mzk5li6nmvs3a

Intrusion detection with evolutionary learning classifier systems

Kamran Shafi, Tim Kovacs, Hussein A. Abbass, Weiping Zhu
2007 Natural Computing  
Evolutionary Learning Classifier Systems (LCSs) combine reinforcement learning or supervised learning with effective genetics-based search techniques.  ...  We detect little sign of overfitting in XCS but somewhat more in UCS. However, both systems tend to reach near-best performance in very few passes over the training data.  ...  Best action maps in XCS As noted in Sect. 3 the major differences between XCS and UCS are that XCS is a reinforcement learner while UCS is supervised, and that XCS learns a complete map while UCS learns  ... 
doi:10.1007/s11047-007-9053-9 fatcat:hwjdqhdph5gq3b3xo3a4apyu2y

Revisiting UCS: Description, Fitness Sharing, and Comparison with XCS [chapter]

Albert Orriols-Puig, Ester Bernadó-Mansilla
2008 Lecture Notes in Computer Science  
This paper provides a deep insight into the learning mechanisms of UCS, a learning classifier system (LCS) derived from XCS that works under a supervised learning scheme.  ...  Besides, we review the fitness computation, based on the individual accuracy of each rule, and introduce a fitness sharing scheme to UCS.  ...  XCS is an accuracy-based learning classifier system introduced by S.W. Wilson [22] .  ... 
doi:10.1007/978-3-540-88138-4_6 fatcat:57dlbeq7k5bx7nzlqs267issdi

Classifier Fitness Based on Accuracy

Stewart W. Wilson
1995 Evolutionary Computation  
These aspects of XCS result in its population tending to form a complete and accurate mapping X x A + P from inputs and actions to payoff predictions.  ...  Besides introducing a new direction for classifier system research, these properties of XCS make it suitable for a wide range of reinforcement learning situations where generalization over states is desirable  ...  If, on the other hand-as in reinforcement learningthe system were oriented toward learning relatively complete maps of the consequences of each action in each niche, then determining the most remunerative  ... 
doi:10.1162/evco.1995.3.2.149 fatcat:h5t5olb5pjbi3nn25oxahl3zcy

Revisit of Rule-Deletion Strategy for XCSAM Classifier System on Classification

Masaya Nakata, Tomoki Hamagami
2017 Transactions of the Institute of Systems Control and Information Engineers  
The XCSAM classifier system is an approach of evolutionary rule-based machine learning, which evolves rules advocating the highest-return actions at state, resulting in best classification.  ...  This paper starts with claiming a limitation that XCSAM still fails to evolutionary generate adequate rules advocating the highest-return actions.  ...  Introduction Learning Classifier System (LCS) [6] is an evolutionary rule-based machine learning method which combines evolutionary computation [5] with a rulebased machine learning such as reinforcement  ... 
doi:10.5687/iscie.30.273 fatcat:eba2jzyvszdabfeqwyllsflp44

Special issue on the 20th anniversary of XCS

Tim Kovacs, Muhammad Iqbal, Kamran Shafi, Ryan Urbanowicz
2015 Evolutionary Intelligence  
The fact that the map is complete allows XCS to learn about every state/action combination, which is an advantage in RL tasks given the typical uncertainty about which action is best for a given state.  ...  One feature of UCS is that it learns only rules whose actions it uses, rather than a complete map, so its rule population is smaller than XCS's.  ... 
doi:10.1007/s12065-015-0131-0 fatcat:2f52bxzpzfd77hz3je26azrtei

State of XCS Classifier System Research [chapter]

Stewart W. Wilson
2000 Lecture Notes in Computer Science  
XCS is a new kind of learning classifier system that differs from the traditional one primarily in its definition of classifier fitness and its relation to contemporary reinforcement learning.  ...  Introduction A classifier system is a learning system that seeks to gain reinforcement from its environment based on an evolving set of condition-action rules called classifiers.  ...  in much faster learning of the payoff map, at least in terms of real actions in E.  ... 
doi:10.1007/3-540-45027-0_3 fatcat:uz5yxxs6mre4pjnxbvhi4selhe

Standard and averaging reinforcement learning in XCS

Pier Luca Lanzi, Daniele Loiacono
2006 Proceedings of the 8th annual conference on Genetic and evolutionary computation - GECCO '06  
This paper investigates reinforcement learning (RL) in XCS.  ...  First, it formally shows that XCS implements a method of generalized RL based on linear approximators, in which the usual input mapping function translates the state-action space into a niche relative  ...  THE XCS CLASSIFIER SYSTEM XCS works as a typical reinforcement learning algorithm though it is based on a rule based representation [11] .  ... 
doi:10.1145/1143997.1144241 dblp:conf/gecco/LanziL06 fatcat:whqmkq4idvgtrlplzadqsqpjcy

Probabilistic co-adaptive brain–computer interfacing

Matthew J Bryan, Stefan A Martin, Willy Cheung, Rajesh P N Rao
2013 Journal of Neural Engineering  
') over brain and environment state, and (2) actions are selected based on entire belief distributions in order to maximize total expected reward; by employing methods from reinforcement learning, the  ...  on the output of classifiers or regression techniques.  ...  The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Science Foundation or the other funding agencies.  ... 
doi:10.1088/1741-2560/10/6/066008 pmid:24140680 fatcat:4xrhgpwhg5aq3nwamfikekkzi4

Reinforcement Learning based Autonomous Vehicle for Exploration and Exploitation of Undiscovered Track

Razin Bin Issa, Md. Saferi Rahman, Modhumonty Das, Monika Barua, Md. Golam Rabiul Alam
2020 2020 International Conference on Information Networking (ICOIN)  
Learning problems that applies model-based ways. • Q value or action value (Q): Q value is almost similar to value.  ...  Hence, reinforcement learning involves learning to take decisions, mapping situations to actions and maximizing reward signals [16] .  ... 
doi:10.1109/icoin48656.2020.9016539 dblp:conf/icoin/IssaRDBA20 fatcat:4obil5k62vclxeyadvffxwejau

How You Act Tells a Lot: Privacy-Leakage Attack on Deep Reinforcement Learning [article]

Xinlei Pan, Weiyao Wang, Xiaoshuai Zhang, Bo Li, Jinfeng Yi, Dawn Song
2019 arXiv   pre-print
As deep reinforcement learning (DRL) has been deployed in a number of real-world systems, such as indoor robot navigation, whether trained DRL policies can leak private information requires in-depth study  ...  To the best of our knowledge, this is the first work to investigate privacy leakage in DRL settings and we show that DRL-based agents do potentially leak privacy-sensitive information from the trained  ...  Learn Classifier Learn a classifier c : D[π j i ] → i. Test: Apply classifier c on target policy π target . Figure 2 : 2 Figure 2: An example of our abstracted grid map.  ... 
arXiv:1904.11082v1 fatcat:7uh3gjuxmfgv3if676dhfxdvp4

SLAM-Safe Planner: Preventing Monocular SLAM Failure using Reinforcement Learning [article]

Vignesh Prasad, Saurabh Singh, Nahas Pareekutty, Balaraman Ravindran, Madhava Krishna
2017 arXiv   pre-print
This paper presents a novel formulation based on Reinforcement Learning (RL) that generates fail safe trajectories wherein the SLAM generated outputs (scene structure and camera motion) do not deviate  ...  Quintessentially, the RL framework successfully learns the otherwise complex relation between motor actions and perceptual inputs that result in trajectories that do not cause failure of SLAM, which are  ...  Reinforcement Learning Reinforcement Learning (RL) is a learning method based on Markov Decision Processes (MDPs) where actions are performed based on the current state of the system and rewards are obtained  ... 
arXiv:1607.07558v4 fatcat:zmcgeopb6zb7dpykgd37jlqz5m

Deep Reinforcement Learning of Volume-guided Progressive View Inpainting for 3D Point Scene Completion from a Single Depth Image [article]

Xiaoguang Han, Zhaoxuan Zhang, Dong Du, Mingdai Yang, Jingming Yu, Pan Pan, Xin Yang, Ligang Liu, Zixiang Xiong, Shuguang Cui
2019 arXiv   pre-print
We present a deep reinforcement learning method of progressive view inpainting for 3D point scene completion under volume guidance, achieving high-quality scene reconstruction from only a single depth  ...  Our approach is end-to-end, consisting of three modules: 3D scene volume reconstruction, 2D depth map inpainting, and multi-view selection for completion.  ...  Acknowledgements We thank the anonymous reviewers for the insightful and constructive comments.This work was funded in part by The Pearl River Talent  ... 
arXiv:1903.04019v2 fatcat:kd6vrz4cyzezfluix2k27loxzq
« Previous Showing results 1 — 15 out of 32,858 results