Rita Chattopadhyay

Multi-source domain adaptation and its application to early detection of fatigue

We consider the characterization of muscle fatigue through noninvasive sensing mechanism such as surface electromyography (SEMG). While changes in the properties of SEMG signals with respect to muscle fatigue have been reported in the literature, the large variation in these signals across different individuals makes the task of modeling and classification of SEMG signals challenging. Indeed, the variation in SEMG parameters from subject to subject creates differences in the data distribution. In this paper, we propose a transfer learning framework based on the multi-source domain adaptation methodology for detecting different stages of fatigue using SEMG signals, that addresses the distribution differences. In the proposed framework, the SEMG data of a subject represent a domain; data from multiple subjects in the training set form the multiple source domains and the test subject data form the target domain. SEMG signals are predominantly different in conditional probability distribution across subjects. The key feature of the proposed framework is a novel weighting scheme that addresses the conditional probability distribution differences across multiple domains (subjects). We have validated the proposed framework on Surface Electromyogram signals collected from 8 people during a fatigue-causing repetitive gripping activity. Comprehensive experiments on the SEMG data set demonstrate that the proposed method improves the classification accuracy by 20% to 30% over the cases without any domain adaptation method and by 13% to 30% over the existing state-of-the-art domain adaptation methods.

Batch mode active sampling based on marginal probability distribution matching

Active Learning is a machine learning and data mining technique that selects the most informative samples for labeling and uses them as training data; it is especially useful when there are large amount of unlabeled data and labeling them is expensive. Recently, batch-mode active learning, where a set of samples are selected concurrently for labeling, based on their collective merit, has attracted a lot of attention. The objective of batch-mode active learning is to select a set of informative samples so that a classifier learned on these samples has good generalization performance on the unlabeled data. Most of the existing batch-mode active learning methodologies try to achieve this by selecting samples based on varied criteria. In this paper we propose a novel criterion which achieves good generalization performance of a classifier by specifically selecting a set of query samples that minimizes the difference in distribution between the labeled and the unlabeled data, after annotation. We explicitly measure this difference based on all candidate subsets of the unlabeled data and select the best subset. The proposed objective is an NP-hard integer programming optimization problem. We provide two optimization techniques to solve this problem. In the first one, the problem is transformed into a convex quadratic programming problem and in the second method the problem is transformed into a linear programming problem. Our empirical studies using publicly available UCI datasets and a biomedical image dataset demonstrate the effectiveness of the proposed approach in comparison with the state-of-the-art batch-mode active learning methods. We also present two extensions of the proposed approach, which incorporate uncertainty of the predicted labels of the unlabeled data and transfer learning in the proposed formulation. Our empirical studies on UCI datasets show that incorporation of uncertainty information improves performance at later iterations while our studies on 20 Newsgroups dataset show that transfer learning improves the performance of the classifier during initial iterations.

Batch Mode Active Sampling Based on Marginal Probability Distribution Matching

Active Learning is a machine learning and data mining technique that selects the most informative samples for labeling and uses them as training data; it is especially useful when there are large amount of unlabeled data and labeling them is expensive. Recently, batch-mode active learning, where a set of samples are selected concurrently for labeling, based on their collective merit, has attracted a lot of attention. The objective of batch-mode active learning is to select a set of informative samples so that a classifier learned on these samples has good generalization performance on the unlabeled data. Most of the existing batch-mode active learning methodologies try to achieve this by selecting samples based on certain criteria. In this article we propose a novel criterion which achieves good generalization performance of a classifier by specifically selecting a set of query samples that minimize the difference in distribution between the labeled and the unlabeled data, after annotation. We explicitly measure this difference based on all candidate subsets of the unlabeled data and select the best subset. The proposed objective is an NP-hard integer programming optimization problem. We provide two optimization techniques to solve this problem. In the first one, the problem is transformed into a convex quadratic programming problem and in the second method the problem is transformed into a linear programming problem. Our empirical studies using publicly available UCI datasets and two biomedical image databases demonstrate the effectiveness of the proposed approach in comparison with the state-of-the-art batch-mode active learning methods. We also present two extensions of the proposed approach, which incorporate uncertainty of the predicted labels of the unlabeled data and transfer learning in the proposed formulation. In addition, we present a joint optimization framework for performing both transfer and active learning simultaneously unlike the existing approaches of learning in two separate stages, that is, typically, transfer learning followed by active learning. We specifically minimize a common objective of reducing distribution difference between the domain adapted source, the queried and labeled samples and the rest of the unlabeled target domain data. Our empirical studies on two biomedical image databases and on a publicly available 20 Newsgroups dataset show that incorporation of uncertainty information and transfer learning further improves the performance of the proposed active learning based classifier. Our empirical studies also show that the proposed transfer-active method based on the joint optimization framework performs significantly better than a framework which implements transfer and active learning in two separate stages.

Towards fatigue and intensity measurement framework during continuous repetitive activities

With the recent advancement in the wearable sensor technology there has been many studies about recognizing users activities, location or environment, but they did not recognize the effect of these activities on the physiological state of the person. The two major physiological aspects associated with any activity are intensity of activity and associated fatigue. Fatigue is an universal human experience that can negatively affect daily life activities. In this paper, we present a framework to measure the level of fatigue and intensity of activity during repetitive daily life activities. The proposed framework acquires and processes time series data from a surface Electromyogram (sEMG) sensor and employs state of art machine learning and data mining techniques to measure the physiological status. We tested this framework using the raw sEMG signals from the hand muscles of 10 subjects, including male and female, of age group around 25 to 45 years, collected during the continuous monitoring of repetitive palm movements at different repetition speeds. The framework graded the levels of fatigue and intensity of activity in a scale of 0 to 1 with an accuracy of 88% with AdaBoost, 94% with SVM, 96% with both HMM and KNN based machine learning techniques.

Subject independent computational framework for myoelectric signals

A subject independent computational framework is one which do not require to be calibrated by the specific subject data to be ready to be used on the subject. The greatest challenge in developing such a framework is the variation in parameters across subjects which is termed as subject based variability. Subject based variability is the variability in data across subjects for the same task, activity or physiological condition. Physiological signals are highly subject specific in nature. Myoelectric signals are one such physiological signals generated in the muscles during any musco-skeletal activity of the body. Spectral and amplitude variations in the myoelectric signals are analyzed to determine the physiological status of a muscle with respect to the intensity of activity and the fatigue state of the muscle. But variations in the spectrum and magnitude of myoelectric signals across subjects pose a great challenge in developing a generalized framework for detecting physiological status of the muscle. In this paper we present statistical tools and techniques to measure subject based variability in myoelectric signals and also present a novel feature selection method based on robustness to subject based variability, with the aim of developing a subject independent measurement framework for fatigue using myoelectric signals. The proposed method provides a subject independent classification accuracy of 80.65%, which is an improvement of 10% to 18% compared to the existing techniques when tested with a wide range of classifiers such SVM, HMM, AdaBoost and KNN. More information and source code are available from the authors.

Processing body sensor data streams for continuous physiological monitoring

With the recent advancement in the wearable sensor technology, various body sensor network systems are being incorporated in the garments to monitor continuous physiological as well as motor behavior of an individual. The raw physiological time series data coming from on-body sensors requires a thorough analysis for extraction of meaningful information. In addition, extracted information need to be presented/recommended to monitoring personnel/self to derive the high-level interpretation of the physiological state without having domain knowledge. In this paper, we propose a knowledge management system that extracts and conveys the information of the physiological states using individualized factor analysis model. The factor analysis based on the quantitative features extracted from the raw data streams provides the hidden knowledge components in the form of latent factors. We tested this system on the raw electromyogram signals from the hand muscles collected during the continuous monitoring of repetitive hand movements, where the hidden information in the form of intensity level of the activity and the muscle fatigue was extracted from the time and frequency domain features.

Optimization-Based Domain Adaptation towards Person-Adaptive Classification Models

The emergence of inexpensive and unobtrusive physiological sensors has widened their application to newer and innovative areas including proactive health monitoring, smart environments and novel human-computer interfaces. The inherent variability in physiological signals across subjects poses a great challenge to traditional machine learning algorithms which are used to develop generalized classification frameworks. In this paper, we propose an optimization-based domain adaptation (ODA) methodology which can provide reliable classification on a given test subject, using the available data from other subjects. The proposed ODA method selects instances from the source domain (data available from other subjects) based on a novel optimization formulation, to ensure that the selected instances are similar in distribution to the target domain (test subject data) in both marginal and conditional probability distributions. We validated the proposed framework on Surface Electromyogram (SEMG) signals collected from 8 people during a fatigue-causing repetitive gripping activity, to detect different stages of fatigue. Comprehensive experiments on our SEMG data set demonstrated that the proposed method improves the classification accuracy by 19% to 21% over traditional classification models, and by 12% to 18% over existing state-of-the-art domain adaptation methodologies.

A Subject-Independent Method for Automatically Grading Electromyographic Features During a Fatiguing Contraction

Many studies have attempted to monitor fatigue from electromyogram (EMG) signals. However, fatigue affects EMG in a subject-specific manner. We present here a subject-independent framework for monitoring the changes in EMG features that accompany muscle fatigue based on principal component analysis and factor analysis. The proposed framework is based on several time- and frequency-domain features, unlike most of the existing work, which is based on two to three features. Results show that latent factors obtained from factor analysis on these features provide a robust and unified framework. This framework learns a model from EMG signals of multiple subjects, that form a reference group, and monitors the changes in EMG features during a sustained submaximal contraction on a test subject on a scale from zero to one. The framework was tested on EMG signals collected from 12 muscles of eight healthy subjects. The distribution of factor scores of the test subject, when mapped onto the framework was similar for both the subject-specific and subject-independent cases.

Hierarchical domain adaptation for SEMG signal classification across multiple subjects

Large variations in Surface Electromyogram (SEMG) signal across different subjects make the process of automated signal classification as a generalized tool, challenging. In this paper, we propose a domain adaptation methodology that addresses this challenge. In particular we propose a hierarchical sample selection methodology, that selects samples from multiple training subjects, based on their similarity with the target subject at different levels of granularity. We have validated our framework on SEMG data collected from 8 people during a fatiguing exercise. Comprehensive experiments conducted in the paper demonstrate that the proposed method improves the subject independent classification accuracy by 21% to 23% over the cases without domain adaptation methods and by 14% to 20% over the existing state-of-the-art domain adaptation methods.