Diptendu Bhattacharya

Secondary factor induced stock index time-series prediction using Self-Adaptive Interval Type-2 Fuzzy Sets

The paper introduces an alternative approach to time-series prediction for stock index data using Interval Type-2 Fuzzy Sets. The work differs from the existing research on time-series prediction by the following counts. First, partitions of the time-series, obtained by fragmenting its valuation space over disjoint equal sized intervals, are represented by Interval Type-2 Fuzzy Sets (or Type-1 fuzzy sets in absence of sufficient data points in the partitions). Second, an Interval Type-2 (or type-1) fuzzy reasoning is performed using prediction rules, extracted from the (main factor) time-series. Third, a type-2 (or type-1) centroidal defuzzification is undertaken to determine crisp measure of inferences obtained from the fired rules, and lastly a weighted averaging of the defuzzified outcomes of the fired rules is performed to predict the time-series at the next time point from its current value. Besides the above three main prediction steps, the other issues considered in the paper include: (i) employing a new strategy to induce the main factor time-series prediction by its secondary factors (other reference time-series) and (ii) self-adaptation of membership functions to properly tune them to capture the sudden changes in the main-factor time-series. Performance analysis undertaken reveals that the proposed prediction algorithm outperforms existing algorithms with respect to root mean-square error by a large margin (?23%). A statistical analysis undertaken with paired t-test confirms that the proposed method is superior in performance at 95% confidence level to most of the existing techniques with root mean square error as the key metric.

Automatic Gesture Recognition for Health Care Using ReliefF and Fuzzy kNN

This work describes a simple method to detect gestures revealing muscle and joint pain. The data is acquired using Kinect Sensor. For the purpose of feature extraction, the twenty joint coordinates are processed in three dimensional space. From each frame, 171 Euclidean distances are calculated and to reduce the dimension of the feature space, ReliefF algorithm is implemented. The classification stage is consists of fuzzy k-nearest neighbour classifier. The proposed method is employed to recognize 24 body gestures and yields a high recognition rate of 90.63 % which is comparatively higher than several other algorithms for young person gesture recognition works.

Self-adaptive type-1/type-2 hybrid fuzzy reasoning techniques for two-factored stock index time-series prediction

Considerable research outcomes on stock index time-series prediction using classical (type-1) fuzzy sets are available in the literature. However, type-1 fuzzy sets cannot fully capture the uncertainty involved in prediction because of its limited representation capability. This paper fills the void. Here, we propose four chronologically improved methods of time-series prediction using interval type-2 fuzzy sets. The first method is concerned with prediction of the (main factor) variation time-series using interval type-2 fuzzy reasoning. The second method considers secondary factor variation as an additional condition in the antecedent of the rules used for prediction. Another important aspect of the first and the second methods is non-uniform partitioning of the dynamic range of the time-series using evolutionary algorithm, so as to ensure that each partition includes at least one data point. The third method considers uniform partitioning without imposing any restriction on the number of data points in a partition. The partitions are here modeled by type-1 fuzzy sets, if there exists a single block of contiguous data, and by interval type-2 fuzzy sets, if there exists two or more blocks of contiguous data in a partition. The fourth method keeps provision for tuning of membership functions using recent data from the given time-series to influence the prediction results with the current trends. Experiments undertaken confirm that the fourth technique outperforms the first three techniques and also the existing techniques with respect to root-mean-square error metric.

Time-Series Prediction and Applications

This book presents machine learning and type-2 fuzzy sets for the prediction of time-series with a particular focus on business forecasting applications. It also proposes new uncertainty management techniques in an economic time-series using type-2 fuzzy sets for prediction of the time-series at a given time point from its preceding value in fluctuating business environments. It employs machine learning to determine repetitively occurring similar structural patterns in the time-series and uses stochastic automaton to predict the most probabilistic structure at a given partition of the time-series. Such predictions help in determining probabilistic moves in a stock index time-series Primarily written for graduate students and researchers in computer science, the book is equally useful for researchers/professionals in business intelligence and stock index prediction. A background of undergraduate level mathematics is presumed, although not mandatory, for most of the sections. Exercises with tips are provided at the end of each chapter to the readers ability and understanding of the topics covered.

Real-time surface material identification using infrared sensor to control speed of an arduino based car like mobile robot

Recently autonomous cars are being in demand to launch into the market for safety and luxury. Autonomous car is an interesting area of research for Engineers and Researchers. Running an automated car on the road in real-time requires several factors to consider. Among them detecting the nature and type of the road is one major aspect. For different types of road surfaces, the control of speed, acceleration, break etc are required to adjust in a proper manner. This paper gives a very cost effective and efficient way of designing a surface identifier for fully autonomous cars. Using the surface identification module the car can identify the surface just in front of it and accordingly it can adjust its speed to move safely.

Music-Induced Emotion Classification from the Prefrontal Hemodynamics

Most of the traditional works on emotion recognition utilize manifestation of emotion in face, voice, gesture/posture and bio-potential signals of the subjects. However, these modalities of emotion recognition cannot totally justify its significance because of wide variations in these parameters due to habitat and culture. The paper aims at recognizing emotion of people directly from their brain response to infrared signal using music as the stimulus. A type-2 fuzzy classifier has been used to eliminate the effect of intra and inter-personal variations in the feature-space, extracted from the infrared response of the brain. A comparative analysis reveals that the proposed interval type-2 fuzzy classifier outperforms its competitors by classification accuracy as the metric.

Handling Main and Secondary Factors in the Antecedent for Type-2 Fuzzy Stock Prediction

Traditional fuzzy logic based approaches to prediction of stock index time-series utilize the reasoning mechanisms of type-1 fuzzy sets. The predictions undertaken thereby occasionally suffer from representational uncertainty. This chapter introduces interval type-2 fuzzy reasoning to capture the uncertainty buried under the individual partitions of a time-series. It presents three different methods of autonomous construction of membership functions, and one additional method for automatic adaptation of membership function for further tuning of memberships with the latest data of the time-series. The first method employs interval type-2 fuzzy reasoning to predict the next day variation in main factor time-series from its current value. The second method too introduces an interval type-2 reasoning with secondary factor variation as an additional antecedent for the prediction. It organizes the dynamic range of the (main factor) time-series as non-uniformly partitioned segments using evolutionary algorithm, so that each partition includes at least one data point sufficient to capture the uncertainty by interval type-2 model. The third method employs uniform partitioning with no restriction on the number of data points in the partitions. It employs type-1 fuzzy sets to capture the uncertainty in a partition when it includes a single block of contiguous data and an interval type-2 fuzzy set when the partition includes two or more blocks. The last method involves additional tuning of the membership functions with recent data from the time-series to imbibe the prediction results with the current trends. Experiments undertaken reveal that the third method with provisions for adaptation of membership functions with recent data outperforms the first two methods. The said method also outperforms existing techniques by a large margin of root mean square error.

An Introduction to Time-Series Prediction

This chapter provides an introduction to time-series prediction. It begins with a formal definition of time-series and gradually explores possible hindrances in predicting a time-series. These hindrances add uncertainty in time-series prediction. To cope up with uncertainty management, the chapter examines the scope of fuzzy sets and logic in the prediction of time-series. Besides dealing with uncertainty, the other important aspect in time-series prediction is to learn the structures embedded in the time-series. The chapter addresses the scope of machine learning in both prediction of the series and also the structures hiding inside the series. The influence of secondary factors in the main-factor time-series is reviewed and possible strategies to utilize secondary factors in predicting main factor time-series are addressed. The methodologies used to partition the dynamic range of a time-series for possible labeling of the diurnal series value in terms of partition number and also for prediction of the next time-point value in terms of the partition number are reviewed, and possible strategies for alternative approaches to partitioning the time-series are overviewed. The chapter ends with a discussion on the scope of the work, highlighting the goals and possible explorations and challenges of economic time-series prediction.

Self-adaptive Interval Type-2 Fuzzy Set Induced Stock Index Prediction

This chapter introduces an alternative approach to time-series prediction for stock index data using Interval Type-2 Fuzzy Sets. The work differs from the existing research on time-series prediction by the following counts. First, partitions of the time-series, obtained by fragmenting its valuation space over disjoint equal sized intervals, are represented by Interval Type-2 Fuzzy Sets (or Type-1 fuzzy sets in absence of sufficient data points in the partitions). Second, an interval type-2 (or type-1) fuzzy reasoning is performed using prediction rules, extracted from the (main factor) time-series. Third, a type-2 (or type-1) centroidal defuzzification is undertaken to determine crisp measure of inferences obtained from the fired rules, and lastly a weighted averaging of the defuzzified outcomes of the fired rules is performed to predict the time-series at the next time point from its current value. Besides the above three main prediction steps, the other issues considered in this chapter include: (i) employing a new strategy to induce the main factor time-series prediction by its secondary factors (other reference time-series), and (ii) self-adaptation of membership functions to properly tune them to capture the sudden changes in the main-factor time-series. Performance analysis undertaken reveals that the proposed prediction algorithm outperforms existing algorithms with respect to root mean-square error by a large margin (≥23%). A statistical analysis undertaken with paired t-test confirms that the proposed method is superior in performance at 95% confidence level to most of the existing techniques with root mean square error as the key metric.

Learning Structures in an Economic Time-Series for Forecasting Applications

The chapter introduces a machine learning approach to knowledge acquisition from a time-series by incorporating three fundamental steps. The first step deals with segmentation of the time-series into time-blocks of non-uniform length with distinguishable characteristics from their neighbours. The second step groups structurally similar time-blocks into clusters by an extension of the DBSCAN algorithm to incorporate multilevel hierarchical clustering. The third step involves representation of the time-series by a special type of automaton with no fixed start or end states. The states in the proposed automaton represent (horizontal) partitions of the time-series, while the cluster centres obtained in the second step are used as input symbols to the states. The state-transitions here are attached with two labels: probability of the transition due the input symbol at the current state and the expected time required for the transition. Once an automaton is built, the knowledge acquisition (training) phase is over. During the test phase, the automaton is consulted to predict the most probable sequence of symbols at a given starting state and the approximate time required (within user-defined margin) to reach a user-defined target state with its probability of occurrence. Test phase prediction accuracy being high over 90%, the proposed prediction can be utilized for trading and investment in stock market.