Optimum-Path Forest

Front Cover
Optimum-Path Forest
Copyright
Dedication
Contents
List of contributors
Biography of the editors
Preface
1 Introduction
	References
2 Theoretical background and related works
	2.1 Introduction
	2.2 The optimum-path forest framework
		2.2.1 Theoretical background
		2.2.2 Supervised learning
			2.2.2.1 OPF using complete graph
			2.2.2.2 OPF using k-nn graph
		2.2.3 Semisupervised learning
		2.2.4 Unsupervised learning
	2.3 Applications
		2.3.1 Supervised
			2.3.1.1 Improvements in training
			2.3.1.2 Improvements in classification
			2.3.1.3 Variations in learning
			2.3.1.4 Biological sciences
			2.3.1.5 Biometrics
			2.3.1.6 Electrical engineering
			2.3.1.7 Geosciences and remote sensing
			2.3.1.8 Image and video analysis
			2.3.1.9 Materials engineering
			2.3.1.10 Medicine
			2.3.1.11 Network security
			2.3.1.12 Feature selection
			2.3.1.13 Petroleum exploration
			2.3.1.14 Other applications
			2.3.1.15 Voice recognition
		2.3.2 Semisupervised
		2.3.3 Unsupervised
			2.3.3.1 Electrical engineering
			2.3.3.2 Image and video processing
			2.3.3.3 Medicine
			2.3.3.4 Network security
			2.3.3.5 Remote sensing images
			2.3.3.6 Other applications
	2.4 Conclusions and future trends
	Acknowledgments
	References
3 Real-time application of OPF-based classifier in Snort IDS
	3.1 Introduction
	3.2 Intrusion detection systems
		3.2.1 Detection approaches in IDS
		3.2.2 Anomaly detection techniques
		3.2.3 Types of IDS
		3.2.4 Open source IDS
			3.2.4.1 Snort
	3.3 Machine learning
		3.3.1 Learning methods
		3.3.2 Algorithms
			3.3.2.1 Optimum-path forest
		3.3.3 Metrics for effectiveness analysis
	3.4 Methodology
		3.4.1 CICIDS2017 data set
		3.4.2 Data set balancing
		3.4.3 ml_classifiers plugin
			3.4.3.1 Network traffic flow management
			3.4.3.2 Classification of network traffic flows
			3.4.3.3 Plugin configuration
	3.5 Experiments and results
		3.5.1 First stage of experiments
			3.5.1.1 Naive Bayes
			3.5.1.2 Decision tree
			3.5.1.3 Random forests
			3.5.1.4 Support vector machine
			3.5.1.5 Optimum-path forest
			3.5.1.6 AdaBoost
			3.5.1.7 Comparison of classification techniques
		3.5.2 Second stage of experiments
			3.5.2.1 DoS slowloris
			3.5.2.2 DoS SlowHTTPTest
			3.5.2.3 DoS hulk
			3.5.2.4 Port scan
			3.5.2.5 SSH brute force
	3.6 Final considerations
		3.6.1 Future works
	Acknowledgments
	References
4 Optimum-path forest and active learning approaches for content-based medical image retrieval
	4.1 Introduction
	4.2 Methodology
		4.2.1 Active learning strategy
	4.3 Experiments
		4.3.1 Results and discussion
	4.4 Conclusion
	4.5 Funding and acknowledgments
	References
5 Hybrid and modified OPFs for intrusion detection systems and large-scale problems
	5.1 Introduction
	5.2 Modified OPF-based IDS using unsupervised learning and social network concept
	5.3 Hybrid IDS using unsupervised OPF based on MapReduce approach
	5.4 Hybrid IDS using modified OPF and selected features
	5.5 Modified OPF using Markov cluster process algorithm
	5.6 Modified OPF based on coreset concept
		5.6.1 Partitioning step
		5.6.2 Sampling step
	5.7 Enhancement of MOPF using k-medoids algorithm
	References
6 Detecting atherosclerotic plaque calcifications of the carotid artery through optimum-path forest
	6.1 Introduction
	6.2 Theoretical background
		6.2.1 Computer-aided diagnosis of atherosclerotic lesions
		6.2.2 Optimum-path forest
			6.2.2.1 Optimum-path forest classifier
			6.2.2.2 Probabilistic optimum-path forest
			6.2.2.3 Optimum-path forest-based approach for anomaly detection
			6.2.2.4 Fuzzy optimum-path forest
			6.2.2.5 Optimum-path forest based on k-connectivity
	6.3 Methodology
		6.3.1 Data set
		6.3.2 Features set
		6.3.3 Metrics
		6.3.4 Experimental setup
	6.4 Experimental results
		6.4.1 Classification
		6.4.2 Statistical analysis
		6.4.3 Computational burden
	6.5 Conclusions and future works
	References
7 Learning to weight similarity measures with Siamese networks: a case study on optimum-path forest
	7.1 Introduction
	7.2 Theoretical background
		7.2.1 Optimum-path forest
			Training step
			Testing step
		7.2.2 Siamese networks
	7.3 Methodology
		7.3.1 Proposed approach
		7.3.2 Data sets
		7.3.3 Experimental setup
	7.4 Experimental results
		7.4.1 BBC News
		7.4.2 Caltech101 Silhouettes
		7.4.3 MPEG-7
		7.4.4 Semeion
	7.5 Conclusion
	References
8 An iterative optimum-path forest framework for clustering
	8.1 Introduction
	8.2 Related work
	8.3 The iterative optimum-path forest framework
		8.3.1 Seed set selection
		8.3.2 Clustering by optimum-path forest
		8.3.3 Seed recomputation
		8.3.4 Returning the forest with lowest total path-cost
		8.3.5 Algorithm outline
		8.3.6 Application to object delineation
	8.4 Experimental results
		8.4.1 Object delineation by iterative dynamic trees
		8.4.2 Analysis on road networks
		8.4.3 Experiments on synthetic data sets
	8.5 Conclusions and future work
	Acknowledgments
	References
9 Future trends in optimum-path forest classification
	References
Index
Back Cover