An electrocardiograph (ECG) is widely used in diagnosis and prediction of cardiovascular diseases (CVDs). The traditional ECG classification methods have complex signal processing phases that leads to expensive designs. This paper provides a deep learning (DL) based system that employs the convolutional neural networks (CNNs) for classification of ECG signals present in PhysioNet MIT-BIH Arrhythmia database. The proposed system implements 1-D convolutional deep residual neural network (ResNet) model that performs feature extraction by directly using the input heartbeats. We have used synthetic minority oversampling technique (SMOTE) that process class-imbalance problem in the training dataset and effectively classifies the five heartbeat types in the test dataset. The classifier’s performance is evaluated with ten-fold cross validation (CV) using accuracy, precision, sensitivity, F1-score, and kappa. We have obtained an average accuracy of 98.63%, precision of 92.86%, sensitivity of 92.41%, and specificity of 99.06%. The average F1-score and Kappa obtained were 92.63% and 95.5% respectively. The study shows that proposed ResNet performs well with deep layers compared to other 1-D CNNs.

Detection and removal of outliers in a dataset is a fundamental preprocessing task without which the analysis of the data can be misleading. Furthermore, the existence of anomalies in the data can heavily degrade the performance of machine learning algorithms. In order to detect the anomalies in a dataset in an unsupervised manner, some novel statistical techniques are proposed in this paper. The proposed techniques are based on statistical methods considering data compactness and other properties. The newly proposed ideas are found efficient in terms of performance, ease of implementation, and computational complexity. Furthermore, two proposed techniques presented in this paper use transformation of data to a unidimensional distance space to detect the outliers, so irrespective of the data’s high dimensions, the techniques remain computationally inexpensive and feasible. Comprehensive performance analysis of the proposed anomaly detection schemes is presented in the paper, and the newly proposed schemes are found better than the state-of-the-art methods when tested on several benchmark datasets.

Numerous advancements in various fields, including pattern recognition and image classification, have been made thanks to modern computer vision and machine learning methods. The capsule network is one of the advanced machine learning algorithms that encodes features based on their hierarchical relationships. Basically, a capsule network is a type of neural network that performs inverse graphics to represent the object in different parts and view the existing relationship between these parts, unlike CNNs, which lose most of the evidence related to spatial location and requires lots of training data. So, we present a comparative review of various capsule network architectures used in various applications. The paper’s main contribution is that it summarizes and explains the significant current published capsule network architectures with their advantages, limitations, modifications, and applications.

Electronic nose is an instrument equipped with chemical gas sensors and is used to sense, identify, and measure different odors. The problem arises when these sensors incorporate drift by the passage of time. The effect of drift is so adverse that the pattern recognition algorithms used for identification and measurement of odors fail to respond accurately. To overcome the challenge of drift in sensors, one of the most widely used techniques is system re-calibration, which is a cumbersome process. Keeping in mind the challenges of drift and issues of system re-calibration for real-life applications, this paper proposes a novel method to compensate drift in gas sensors with the following contributions: 1) the fitness function of a recursive metaheuristic optimization method is modified by embedding random forests learning for the quantification of six different gases under drift; 2) the proposed approach is able to compensate the long-term sensors drift without requiring any system re-calibration; and 3) the feature vector exploitation using particle swarm optimization reduces the computational complexity and increases the prediction accuracy of the system. A comparison is provided with different state-of-the-art approaches, and the proposed approach is found better in terms of prediction accuracy when tested on a benchmark dataset publically available.

The video classification task has gained significant success in the recent years. Specifically, the topic has gained more attention after the emergence of deep learning models as a successful tool for automatically classifying videos. In recognition of the importance of the video classification task and to summarize the success of deep learning models for this task, this paper presents a very comprehensive and concise review on the topic. There are several existing reviews and survey papers related to video classification in the scientific literature. However, the existing review papers do not include the recent state-of-art works, and they also have some limitations. To provide an updated and concise review, this paper highlights the key findings based on the existing deep learning models. The key findings are also discussed in a way to provide future research directions. This review mainly focuses on the type of network architecture used, the evaluation criteria to measure the success, and the datasets used. To make the review self-contained, the emergence of deep learning methods towards automatic video classification and the state-of-art deep learning methods are well explained and summarized. Moreover, a clear insight of the newly developed deep learning architectures and the traditional approaches is provided. The critical challenges based on the benchmarks are highlighted for evaluating the technical progress of these methods. The paper also summarizes the benchmark datasets and the performance evaluation matrices for video classification. Based on the compact, complete, and concise review, the paper proposes new research directions to solve the challenging video classification problem.

Long term sensors drift is a challenging problem to solve for instruments like an Electronic Nose System (ENS). These electronic instruments rely on Machine Learning (ML) algorithms for recognizing the sensed odors. The effect of long-term drift influences the performance of ML algorithms and the models those are trained on drift free data fail to perform on the drifted data. Moreover, the response of an electronic nose system depends on the variable response of the sensors and a delay is expected in reaching a steady state by the sensors. In this paper, these two problems of ‘sensors long term drift’ and ‘delayed response’ are solved simultaneously to propose a robust and fast electronic nose system, with following merits: (i) only initial transient state features are used in the proposed system without waiting for the sensors to reach a steady state, (ii) a modified boxplot approach is used to handle noisy/drifted data points as a preprocessing step before the classification setup, (iii) a heuristic tree classification approach with optimized transient features is proposed, (iv) the proposed approach only relies on adapted ML methods contrary to the traditional approaches like system recalibration or sensors replacement for handling sensors drift, and (v) the proposed ML model does not require any target domain data and uses only the source domain data for learning the classifier, opposed to the other ML solutions available in the existing literature. The proposed method is tested using a large scale gas sensors drift benchmark dataset available freely on UCI Machine Learning repository and is found better than the existing state-of-the art approaches with an overall accuracy of 87.34%.