Visualization of learning in multilayer perceptron networks using principal component analysis

Gallagher, Marcus; Downs, T.

doi:10.1109/tsmcb.2003.808183

Cited by 38 publications

(19 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Visualization is an important part of understanding the inner workings of many systems, but particularly those of learning systems [2,6,11,16,17,30]. This paper focuses on visualizing reward properties to aid in both agent reward evaluation and design.…”

Section: Reward Visualizationmentioning

confidence: 99%

Analyzing and visualizing multiagent rewards in dynamic and stochastic domains

Agogino

Tumer

2008

Auton Agent Multi-Agent Syst

View full text Add to dashboard Cite

The ability to analyze the effectiveness of agent reward structures is critical to the successful design of multiagent learning algorithms. Though final system performance is the best indicator of the suitability of a given reward structure, it is often preferable to analyze the reward properties that lead to good system behavior (i.e., properties promoting coordination among the agents and providing agents with strong signal to noise ratios). This step is particularly helpful in continuous, dynamic, stochastic domains ill-suited to simple table backup schemes commonly used in TD(λ)/Q-learning where the effectiveness of the reward structure is difficult to distinguish from the effectiveness of the chosen learning algorithm. In this paper, we present a new reward evaluation method that provides a visualization of the tradeoff between the level of coordination among the agents and the difficulty of the learning problem each agent faces. This method is independent of the learning algorithm and is only a function of the problem domain and the agents' reward structure. We use this reward property visualization method to determine an effective reward without performing extensive simulations. We then test this method in both a static and a dynamic multi-rover learning domain where the agents have continuous state spaces and take noisy actions (e.g., the agents' movement decisions are not always carried out properly). Our results show that in the more difficult dynamic domain, the reward efficiency visualization method provides a two order of magnitude speedup in selecting good rewards, compared to running a full simulation. In addition, this method facilitates the design and analysis of new rewards tailored to the observational limitations of the domain, providing rewards that combine the best properties of traditional rewards.

show abstract

Section: Reward Visualizationmentioning

confidence: 99%

Analyzing and visualizing multiagent rewards in dynamic and stochastic domains

Agogino

Tumer

2008

Auton Agent Multi-Agent Syst

View full text Add to dashboard Cite

show abstract

“…The Multilayer Perceptron (MLP) is the most widely applied and researched artificial neural network (ANN) model. MLP networks implement mappings from input space to output space and are normally applied to supervised learning tasks [24]. The Sigmoidal function was selected as the MLP activation function, with a range of values in the interval [0, 1].…”

Section: B Second Phase Of the Mechanism Of Classificationmentioning

confidence: 99%

A hybrid agent-based classification mechanism to detect denial of service attacks

Trejos¹,

Santana²,

González³

et al. 2009

jopha

View full text Add to dashboard Cite

Abstract-This paper presents the core component of a solution based on agent technology specifically adapted for the classification of SOAP messages. The messages can carry out attacks that target the applications providing Web Services. One of the most common attacks requiring novel solutions is the denial of service attack (DoS), caused for the modifications introduced in the XML of the SOAP messages. The specifications of existing security standards do not focus on this type of attack. This article presents an advanced mechanism of classification designed in two phases incorporated within a CBR-BDI Agent type. This mechanism classifies the incoming SOAP message and blocks the malicious SOAP messages. Its main feature involves the use of decision trees, fuzzy logic rules and neural networks for filtering attacks. These techniques provide a mechanism of classification with the self-adaption ability to the changes that occur in the patterns of attack. A prototype was developed and the results obtained are presented in this study.

show abstract

“…Using the same approach to generate the second direction results in several vertices that lie very close after the projection. Optimization of the average spacing leads to W (2) = (-0.1, 0.2, -0.5, -0.33, 0.75), with all vertices clearly separated (Fig. 1).…”

Section: How To Visualize?mentioning

confidence: 99%

“…They are not helpful to see what type of internal representations have developed for a given data set, or how well the network performs on some data. A few attempts to visualize the training process are restricted to network trajectories in the weight space ( [2], Kordos and Duch, in print). The usefulness of visualization of network outputs has been shown recently [3].…”

Section: Introductionmentioning

confidence: 99%

Visualization of Hidden Node Activity in Neural Networks: II. Application to RBF Networks

Duch

2004

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. Quality of neural network mappings may be evaluated by visual inspection of hidden and output node activities for the training dataset. This paper discusses how to visualize such multidimensional data, introducing a new projection on a lattice of hypercube nodes. It also discusses what type of information one may expect from visualization of the activity of hidden and output layers. Detailed analysis of the activity of RBF hidden nodes using this type of visualization is presented in the companion paper.

show abstract

Visualization of learning in multilayer perceptron networks using principal component analysis

Cited by 38 publications

References 15 publications

Analyzing and visualizing multiagent rewards in dynamic and stochastic domains

Analyzing and visualizing multiagent rewards in dynamic and stochastic domains

A hybrid agent-based classification mechanism to detect denial of service attacks

Visualization of Hidden Node Activity in Neural Networks: II. Application to RBF Networks

Contact Info

Product

Resources

About