Exploiting unlabeled data to improve peer-to-peer traffic classification using incremental tri-training method

Two birds with one stone: Classifying positive and unlabeled examples on uncertain data streams

Identification of network traffic using port-based or payload-based analysis is becoming increasing difficult with many Peer-to-Peer (P2P) application using dynamic ports, masquerading techniques, and encryption to avoid detection. To overcome this problem, several machine learning technique were proposed to classify P2P traffics. But in the real P2P network environment, new communities of peers often attend and old communities of peers often leave. It requires the identification methods to be capable of coping with concept drift, and updating the model incrementally. In this paper, we present a concept-adapting algorithm CluMC which is based on streaming data mining techniques to identify P2P applications in Internet traffic. The CluMC use micro-cluster structures which contain potential micro-cluster structures and outlier micro-cluster structures to classify the P2P traffic and discover the concept drift with limited memory. Our performance study over a number of real data sets that we captured at a main gateway router demonstrates the effectiveness and efficiency of our method. DOI: http://dx.doi.org/10.11591/telkomnika.v11i8.3030

In supervised learning, the learning algorithm uses labeled training examples from every class to generate a classification function. One of the drawbacks of this classic paradigm is that a large number of labeled examples are needed in order to learn accurately. Since labeling is often done manually, it can be very labor intensive and time consuming. In this chapter, we study two partially supervised learning tasks. As their names suggest, these two learning tasks do not need full supervision, and thus are able to reduce the labeling effort. The first is the task of learning from labeled and unlabeled examples, which is commonly known as semisupervised learning. In this chapter, we also call it LU learning (L and U stand for “labeled” and “unlabeled” respectively). In this learning setting, there is a small set of labeled examples of every class, and a large set of unlabeled examples. The objective is to make use of the unlabeled examples to improve learning.

Facing to the high dimensional data, how to deal them well is the most difficult problem in the field of machine learning, pattern recognition and the relative fields. In this paper, we propose a new semi-supervised multi-view learning method, which partition or select the abundant attributes (called attribute partition or attribute selection) into subsets. We consider each subset as a view and on each subset train a classifier to label the unlabeled examples. Based on the ensemble learning, we combine their predictions to classify the unlabeled examples. The semi-supervised learning idea is that to make use of the large number unlabeled example to modify the classifiers iteratively. Experiments on UCI datasets show that this method is feasible and can improve the efficiency. Both theoretical analysis and experiments show that the proposed method has excellent accuracy and speed of classification.

Traditional linear discriminant analysis (LDA) is a popular dimensionality reduction method which preserve class separability. The method needs the labeled data to train. However in real worlds, the labeled training examples are very few but there are sufficient unlabeled data examples, so some former work (SDA) has made use the unlabeled training examples to do dimensionality reduction. But sometimes, there exits another kind of domain knowledge: the instances level constraints. In this paper, we consider the case when there are some useful instance level constraints. We propose a novel algorithm semi-supervised discriminant analyze with constraints (SDAC) which use three kinds of data: very few labeled data examples, sufficient unlabeled data examples and the instance level constraints. Our algorithm can be viewed as a constraint extension of traditional SDA algorithm. Experiments have been presented for semi-supervised classification tasks and have shown the effectiveness of our algorithm.

Huge amount of manual efforts are required to annotate large image/video archives with text annotations. Several recent works attempted to automate this task by employing supervised learning approaches to associate visual information extracted in segmented images with semantic concepts provided by associated text. The main limitation of such approaches, however, is that large labeled training corpus is still needed for effective learning, and semantically meaningful segmentation for images is in general unavailable. This paper explores the use of bootstrapping approach to tackle this problem. The idea is to start from a small set of labeled training examples, and successively annotate a larger set of unlabeled examples. This is done using the cotraining approach, in which two statistically classifiers are used to co-train and co-annotate the unlabeled examples. An active learning approach is used to select the best examples to label at each stage of learning in order to maximize the learning objective. To accomplish this, we break the task of annotating images into the sub-tasks of: (a) segmenting images into meaningful units, (b) extracting appropriate features for the units, and (c) associating these features with text. Because of the uncertainty in sub-tasks (a) and (b), we adopt two independent segmentation methods (task a) and two independent sets of features (task b) to support co-training. We carried out experiments using a mid-sized image collection (comprising about 6,000 images from CorelCD, PhotoCD and Web) and demonstrated that our bootstrapping approach significantly improve the performance of annotation by about 10% in terms of F1 measure as compared to the best results obtained from the traditional supervised learning approach. Moreover, the bootstrapping approach has the key advantage of requiring much fewer labeled examples in training.

Due to the wide proliferation of the Internet and telecommunication, huge amount of information has been produced as digital data format. It is impossible to classify this information with one's own hand one by one in many realistic problems, so that the research on automatic text classification has been grown. Machine learning technologies have applied in text classification. However, the traditional statistic machine learning technologies require large number of labeled training examples to learn accurately. To obtain enough training examples, we have to label on these huge training examples by hand. This paper presents a supervised learning algorithm based on support vector machine (SVM) to classify text documents more accurately by using unlabeled documents to augment available labeled training examples. Experimental results indicate that the classification with unlabeled examples using SVM is superior to the conventional classification,with labeled examples.

Recent research shows that rule based models perform well while classifying large data sets such as data streams with concept drifts. A genetic algorithm is a strong rule based classification algorithm which is used only for mining static small data sets. If the genetic algorithm can be made scalable and adaptable by reducing its I/O intensity, it will become an efficient and effective tool for mining large data sets like data streams. In this paper a scalable and adaptable online genetic algorithm is proposed to mine classification rules for the data streams with concept drifts. Since the data streams are generated continuously in a rapid rate, the proposed method does not use a fixed static data set for fitness calculation. Instead, it extracts a small snapshot of the training example from the current part of data stream whenever data is required for the fitness calculation. The proposed method also builds rules for all the classes separately in a parallel independent iterative manner. This makes the proposed method scalable to the data streams and also adaptable to the concept drifts that occur in the data stream in a fast and more natural way without storing the whole stream or a part of the stream in a compressed form as done by the other rule based algorithms. The results of the proposed method are comparable with the other standard methods which are used for mining the data streams.

An innovative approach for real-time network traffic classification

A word sense disambiguator that is able to distinguish among the many senses of common words that are found in general-purpose, broad-coverage lexicons would be useful. For example, experiments have shown that, given accurate sense disambiguation, the lexical relations encoded in lexicons such as WordNet can be exploited to improve the effectiveness of information retrieval systems. This paper describes a classifier whose accuracy may be sufficient for such a purpose. The classifier combines the output of a neural network that learns topical context with the output of a network that learns local context to distinguish among the senses of highly ambiguous words.The accuracy of the classifier is tested on three words, the noun line, the verb serve, and the adjective hard; the classifier has an average accuracy of 87%, 90%, and 81%, respectively, when forced to choose a sense for all test cases. When the classifier is not forced to choose a sense and is trained on a subset of the available senses, it rejects test cases containing unknown senses as well as test cases it would misclassify if forced to select a sense. Finally, when there are few labeled training examples available, we describe an extension of our training method that uses information extracted from unlabeled examples to improve classification accuracy.

To address the critical industry pain points in streaming data scenarios across industrial production, intelligent transportation and other related fields—namely continuous performance degradation of classification models induced by concept drift, the absence of dedicated closed-loop drift adaptation processing capability in mainstream stream processing systems, poor engineering implementability of specialized tools, and the inherent difficulty in balancing real-time performance and classification accuracy—this paper designs and implements a Dynamic Adaptive Concept Drift Processing System (DACPS) for streaming data classification. The system adopts a hierarchical modular architecture, integrating five core functional modules: streaming data preprocessing, real-time concept drift detection, dynamic sample optimization, incremental model training, and full-process management and control. This architecture enables fully automated and adaptive processing for the end-to-end workflow of streaming data classification and prediction. Extensive multi-dimensional functional tests, specialized performance tests, and scenario-based validation demonstrate that the system can stably adapt to diverse types of concept drift scenarios, and meets all design specifications in terms of classification accuracy, convergence speed, operating efficiency, and resource occupation control. Meanwhile, the system is equipped with an accessible visual interactive interface, making it widely adaptable to the streaming data classification and processing requirements across multiple application domains.

Active reducing classification error for CAD systems

Unlabeled patterns to tighten Rademacher complexity error bounds for kernel classifiers

Aim/Purpose: This paper describes how to use a multilayer perceptron to improve concept drift recovery in streaming environments. Background: Classifying instances in a data stream environment with concept drift is a challenging topic. The base learner must be adapted online to the current data. Several data mining algorithms have been adapted/used to this type of environment. In this study, two techniques are used to speed up the adaptation of an artificial neural network to the current data, increasing its predictive accuracy while detecting the concept drift sooner. Methodology: Experiments were performed to analyze how some techniques behave in different scenarios and compare them with other classifiers built to deal with data streams and concept drifts. Contribution: This study suggests two techniques to improve the classification results: an embedded concept drift detection method to identify when a change has occurred and setting the learning rate to a higher level whenever a new concept is being learned to give more weight to recent instances, with its value decreased over time. Findings: Results indicate that gradually reducing the learning rate with an embedded concept drift detector has better statistical results than other single classifiers built to deal with data streams and concept drifts. Recommendations for Practitioners: Based on the empirical results, this study provides recommendations on how to improve the multilayer perceptron in data stream environments suffering from concept drifts. Recommendation for Researchers: Researchers should conduct investigations to increase the number of base classifiers used in data stream environments and in situations where concept drifts occur. Impact on Society: The objective of this study is to increase the use of multilayer perceptrons in data stream environments suffering from concept drifts, as nowadays, Hoeffding Trees and Naive Bayes are the base classifiers mostly used. Future Research: Additional research includes adapting the online learning rate by increasing/decreasing it based on the performance of the Multilayer Perceptron. This scheme would allow the removal of parameters that must be set by the user, like learning rate upper bound and number of instances to return to the stable value.

Previous research in active learning has primarily focused on selecting examples from closed-set data, which consists solely of unlabeled examples from the target classes. However, this approach overlooks the more prevalent scenario of open-set data in real-world applications. Open-set data encompasses examples from both target classes and non-target classes. To fill this gap, we propose a novel framework called JOSAL, which enhances the accuracy of the classifier by precisely selecting the target class examples from open-set data. The JOSAL framework introduces the concept of joint learning, where the Sampler and Classifier components perform sampling and classification tasks, respectively, by sharing example features extracted from a pre-trained Encoder. To maximize the classification accuracy of the Classifier, the framework adopts a novel joint learning strategy. This strategy initially prioritizes optimizing the Sampler and gradually shifts the optimization attention to the Classifier. The experimental results demonstrate that, compared to baselines, our approach exhibits stronger sampling precision and achieves higher classification accuracy. To the best of our knowledge, this is the first work to address the open-set active learning problem using the joint learning paradigm.