Rule extraction from support vector machines: A review

Barakat, Nahla; Bradley, Andrew P.

doi:10.1016/j.neucom.2010.02.016

Cited by 200 publications

(112 citation statements)

References 38 publications

Supporting

Mentioning

110

Contrasting

Unclassified

Order By: Relevance

“…Datasets and GFA code are available online. 3 Synthetic data. We generated 6, 000 items with 3, 000 assigned to each of two classes.…”

Section: Methodsmentioning

confidence: 99%

“…The first step of the direct influence audit method we will use is the creation of an interpretable model of the model. By a "model of a model" we mean that we should 1) train the model f on training data (X, Y ), 2) determine new labelsŶ from the predicted outcomes f (X), and 3) overfit an interpretable model I(f ) to these predicted labels (as done in [3]). (This idea is similar to model compression, but without the need to find new test data [5].)…”

Section: Evaluating With Respect To a Direct Influence Auditmentioning

confidence: 99%

See 1 more Smart Citation

Auditing black-box models for indirect influence

et al. 2017

View full text Add to dashboard Cite

Abstract-Data-trained predictive models see widespread use, but for the most part they are used as black boxes which output a prediction or score. It is therefore hard to acquire a deeper understanding of model behavior, and in particular how different features influence the model prediction. This is important when interpreting the behavior of complex models, or asserting that certain problematic attributes (like race or gender) are not unduly influencing decisions.In this paper, we present a technique for auditing black-box models, which lets us study the extent to which existing models take advantage of particular features in the dataset, without knowing how the models work. Our work focuses on the problem of indirect influence: how some features might indirectly influence outcomes via other, related features. As a result, we can find attribute influences even in cases where, upon further direct examination of the model, the attribute is not referred to by the model at all.Our approach does not require the black-box model to be retrained. This is important if (for example) the model is only accessible via an API, and contrasts our work with other methods that investigate feature influence like feature selection. We present experimental evidence for the effectiveness of our procedure using a variety of publicly available datasets and models. We also validate our procedure using techniques from interpretable learning and feature selection, as well as against other black-box auditing procedures.

show abstract

“…Datasets and GFA code are available online. 3 Synthetic data. We generated 6, 000 items with 3, 000 assigned to each of two classes.…”

Section: Methodsmentioning

confidence: 99%

Section: Evaluating With Respect To a Direct Influence Auditmentioning

confidence: 99%

Auditing black-box models for indirect influence

et al. 2017

View full text Add to dashboard Cite

show abstract

“…[22], [23], [24]. The SOAR algorithm shown in Figure 1, is independent of the neural structure and here uses a standard Multi-Layer Perceptron (MLP) as a Neural "Oracle" which was chosen as a classifier to produce good generalisation and noise-tolerance.…”

Section: Soar Extractionmentioning

confidence: 99%

Neural Network Rule Extraction to Detect Credit Card Fraud

Ryman-Tubb

Krause

2011

Engineering Applications of Neural Networks

View full text Add to dashboard Cite

Abstract. Neural networks have represented a serious barrier-to-entry in their application in automated fraud detection due to their black box and often proprietary nature which is overcome here by combining them with symbolic rule extraction. A Sparse Oracle-based Adaptive Rule extraction algorithm is used to produce comprehensible rules from a neural network to aid the detection of credit card fraud. In this paper, a method to improve this extraction algorithm is presented along with results from a large real-world European credit card data set. Through this application it is shown that neural networks can assist in mission-critical areas of business and are an important tool in the transparent detection of fraud.

show abstract

“…Knowledge discovery has been successfully used in various application areas: engineering [1], education [2], business and finance [3], insurance [4], telecommunication [5], chemistry [6], and medicine [7]. There are many techniques which are used for knowledge extraction from databases such as neural networks [8][9], genetic algorithms [10][11], decision tree [12][13], instance-based learning [14][15], rule induction [16][17], and support vector machine [18][19]. However, the ANN is still one of the most widely used techniques for knowledge extraction due to their advantages compared with the other techniques.…”

Section: Introductionmentioning

confidence: 99%

Destructive Algorithm for Rule Extraction based on a Trained Neural Network

ElAlami¹

2012

IJCA

View full text Add to dashboard Cite

The present paper introduces a new destructive algorithm for rule extraction based on a trained neural network. The degree of complexity of neural network increases exponentially as a factor of the numbers of input and hidden nodes. Therefore, the dimensionality of the trained neural network is reduced by using a proposed destructive algorithm to extract only the most effective values of the input attributes which have higher impact on the output result for each class. Thus, the searching efficiency is highly increased and the computation is dramatically reduced for extracting rules. The generated rules from the proposed model are fired through two levels for each class. As for the first level, it deals with each individual effective input value, and the second level is concerned with each possible conjunction of the effective input values. Moreover, the proposed model extracts the strongest rules which represent a large number of instances from the database by adjusting the similarity measure threshold value. Finally, the proposed model is evaluated on different public-domain datasets and compared with standard learning models from WEKA, then the results assert that the set of rules extraction from the proposed method is more accurate and concise compared with those obtained by the other models.

show abstract

Rule extraction from support vector machines: A review

Cited by 200 publications

References 38 publications

Auditing black-box models for indirect influence

Auditing black-box models for indirect influence

Neural Network Rule Extraction to Detect Credit Card Fraud

Destructive Algorithm for Rule Extraction based on a Trained Neural Network

Contact Info

Product

Resources

About