Mikhail Dashevskiy

Reliable Probabilistic Classification and Its Application to Internet Traffic

Many machine learning algorithms have been used to classify network traffic flows with good performance, but without information about the reliability in classifications. In this paper, we present a recently developed algorithmic framework, namely the Venn Probability Machine, for making reliable decisions under uncertainty. Experiments on publicly available real traffic datasets show the algorithmic framework works well. Comparison is also made to the published results.

Network Traffic Demand Prediction with Confidence

Many network resource management solutions typically employ traffic prediction algorithms to improve the performance of a network. In this paper we extend a newly developed method for prediction with confidence to time series data and apply it to the network traffic demand prediction problem. We investigate the performance of the proposed algorithm on a number of publicly available network traffic demand datasets. The experimental results are very promising.

Time series prediction with performance guarantee

Time series prediction has many important real applications such as network resource management and quality-of-service assurance. Many different techniques have been developed to deal with time series predictions, for example, the Box–Jenkins approach and machine learning. In this study, the authors focus on the problem of time series prediction with performance guarantees and describe two machine-learning techniques, namely prediction with expert advice and conformal predictors. The authors investigate the application of these techniques to network traffic demand and propose a novel way of combining these two techniques to provide performance guarantee on predictions. The method is generic and the authors demonstrate this approach by carrying out extensive experiments on both artificially generated data and publicly available network traffic demand datasets. Empirical results show that the proposed method can increase the performance of the prediction system.

Predictions with Confidence in Applications

Many applications require predictions with confidence. We are interested in Confidence Machines which are algorithms that can provide some measure on how confident they are that their output is correct. Confidence Machines are quite general and there are many algorithms solving the problem of prediction with confidence. As predictors we consider Venn Probability Machines and Conformal Predictors. Both of these algorithms rely on an underlying algorithm for prediction and in this paper we use two simple algorithms, namely the Nearest Neighbours and Nearest Centroid algorithms. Our aim is to provide some guidelines on how to choose the most suitable algorithm for a practical application where confidence is needed.

RELIABLE PROBABILISTIC CLASSIFICATION OF INTERNET TRAFFIC

Classification of Internet traffic is very important to many applications such as network resource management, network security enforcement and intrusion detection. Many machine-learning algorithms have been successfully used to classify network traffic flows with good performance, but without information about the reliability in classifications. In this paper, we present a recently developed algorithmic framework, namely the Venn Probability Machine, for making reliable decisions under uncertainty. Experiments on publicly available real Internet traffic datasets show the algorithmic framework works well. Comparison is also made to the published results.

Prediction of long-range dependent time series data with performance guarantee

Modelling and predicting long-range dependent time series data can find important and practical applications in many areas such as telecommunications and finance. In this paper, we consider Fractional Autoregressive Integrated Moving Average (FARIMA) processes which provide a unified approach to characterising both short-range and long-range dependence. We compare two linear prediction methods for predicting observations of FARIMA processes, namely the Innovations Algorithm and Kalman Filter, from the computational complexity and prediction performance point of view. We also study the problem of Prediction with Expert Advice for FARIMA and propose a simple but effective way to improve the prediction performance. Alongside the main experts (FARIMA models) we propose to use some naive methods (such as Least-Squares Regression) in order to improve the performance of the system. Experiments on publicly available datasets show that this construction can lead to great improvements of the prediction system.We also compare our approach with a traditional method of model selection for the FARIMA model, namely Akaike Information Criterion.

Guaranteed Network Traffic Demand Prediction Using FARIMA Models

The Fractional Auto-Regressive Integrated Moving Average (FARIMA) model is often used to model and predict network traffic demand which exhibits both long-range and short-range dependence. However, finding the best model to fit a given set of observations and achieving good performance is still an open problem. We present a strategy, namely Aggregating Algorithm, which uses several FARIMA models and then aggregates their outputs to achieve a guaranteed (in a sense) performance. Our feasibility study experiments on the public datasets demonstrate that using the Aggregating Algorithm with FARIMA models is a useful tool in predicting network traffic demand.

Network Traffic Classification and Demand Prediction

Reliable classification of network traffic and accurate demand prediction can offer substantial benefits to service differentiation, enforcement of security policies, and traffic engineering for network operators and service providers. For example, dynamic resource allocation with the support of traffic prediction can efficiently utilize the network resources and support quality of service. One of the key requirements for dynamic resource allocation framework is to predict traffic in the next control time interval based on historical data and online measurements of traffic characteristics over appropriate timescales. Predictions with reliability measures allow service providers and network carriers to effectively perform a cost-benefit evaluation of alternative actions and optimize network performance such as delay and information loss. In this chapter, we apply conformal predictions to two important problems of the network resource management. First, we discuss reliable network traffic classification using network traffic flow measurement. Second, we consider the problem of time series analysis of network resource demand and investigate how to make predictions and build effective prediction intervals. Experimental results on publicly available datasets are be presented to demonstrate benefits of the conformal predictions.

Aggregating Algorithm for a Space of Analytic Functions

In this paper the problem of Prediction with Expert Advice is considered. We apply an existing algorithm, the Aggregating Algorithm, to a specific class of experts. This class of experts approximates (with respect to its parameter) the class of continuous functions and in this way it is close to a natural way of describing a possible dependence between two variables (continuous). We develop an explicit algorithm and prove an upper bound on the difference between the loss of our algorithm and the loss of the best expert, which has the order of the squared logarithm of the number of steps of the algorithm. This bound lies between existing bounds which have the form of the logarithm of the number of steps and the square root of the number of steps. Having more sets (algorithm, class of experts, upper bound) helps in choosing an appropriate way of solving the problem of Prediction with Expert Advice for a particular application.