Papers by Siddhartha Bhattacharyya
A brief review of portfolio optimization techniques
Artificial Intelligence Review
Automatic clustering of colour images using quantum inspired meta-heuristic algorithms
Applied Intelligence
Quantum inspired meta‐heuristic approaches for automatic clustering of colour images
International Journal of Intelligent Systems
In this article, quantum inspired incarnations of two swarm based meta‐heuristic algorithms, name... more In this article, quantum inspired incarnations of two swarm based meta‐heuristic algorithms, namely, Crow Search Optimization Algorithm and Intelligent Crow Search Optimization Algorithm have been proposed for automatic clustering of colour images. The performance and effectiveness of the proposed algorithms have been judged by experimenting on 15 Berkeley images and five publicly available real life images of different sizes. The validity of the proposed algorithms has been justified with the help of four different cluster validity indices, namely, Pakhira Bandyopadhyay Maulik, I‐index, Silhouette and CS‐measure. Moreover, Sobol's sensitivity analysis has been performed to tune the parameters of the proposed algorithms. The experimental results prove the superiority of proposed algorithms with respect to optimal fitness, computational time, convergence rate, accuracy, robustness, t ‐test and Friedman test. Finally, the efficacy of the proposed algorithms has been proved with the help of quantitative evaluation of segmentation evaluation metrics.
Qutrit-Based Genetic Algorithm for Hyperspectral Image Thresholding
Recent Trends in Signal and Image Processing
Informatics and Automation, 2022
The task of reducing the computational complexity of contour detection in images is considered in... more The task of reducing the computational complexity of contour detection in images is considered in the article. The solution to the task is achieved by modifying the Canny detector and reducing the number of passes through the original image. In the first case, two passes are excluded when determining the adjacency of the central pixel with eight adjacent ones in a frame of size 3х3. In the second case, three passes are excluded, two as in the first case and the third one necessary to determine the angle of gradient direction. This passage is provided by a combination of fuzzy rules. The goal of the work is to increase the performance of computational operations in the process of detecting the edges of objects by reducing the number of passes through the original image. The process of edge detection is carried out by some computational operations of the Canny detector with the replacement of the most complex procedures. In the proposed methods, fuzzification of eight input variables ...
The paper describes a self supervised parallel self organizing neural network (PSONN) architectur... more The paper describes a self supervised parallel self organizing neural network (PSONN) architecture for true color image segmentation. The proposed architecture is a parallel extension of the standard single self organizing neural network architecture (SONN) and comprises an input (source) layer of image information, three single self organizing neural network architectures for segmentation of the different primary color components in a color image scene and one final output (sink) layer for fusion of the segmented color component images. Responses to the different shades of color components are induced in each of the three single network architectures (meant for component level processing) by applying a multilevel version of the characteristic activation function, which maps the input color information into different shades of color components, thereby yielding a processed component color image segmented on the basis of the different shades of component colors. The number of target ...
A multilayer self organizing neural neural network (MLSONN) architecture for binary object extrac... more A multilayer self organizing neural neural network (MLSONN) architecture for binary object extraction, guided by a beta activation function and characterized by backpropagation of errors estimated from the linear indices of fuzziness of the network output states, is discussed. Since the MLSONN architecture is designed to operate in a single point fixed/uniform thresholding scenario, it does not take into cognizance the heterogeneity of image information in the extraction process. The performance of the MLSONN architecture with representative values of the threshold parameters of the beta activation function employed is also studied. A three layer bidirectional self organizing neural network (BDSONN) architecture comprising fully connected neurons, for the extraction of objects from a noisy background and capable of incorporating the underlying image context heterogeneity through variable and adaptive thresholding, is proposed in this article. The input layer of the network architect...
Comparative Sentiment Analysis on a Set of Movie Reviews Using Deep Learning Approach
The International Conference on Advanced Machine Learning Technologies and Applications (AMLTA2018), 2018
This paper provides an insight to one of the recent additions in the turf of Machine Learning cul... more This paper provides an insight to one of the recent additions in the turf of Machine Learning culture - the process of learning representation or features, known as Deep Learning. It is highly anticipated that Deep Learning will fare much better than the traditional machine learning algorithms not only because of scalability but also of its ability to perform automatic feature extraction from raw data. This paper deals with the analyzing of sentiments on a set of movie reviews, which is considered to be the most demanding facet of NLP’s. In this paper, Google’s algorithm Word2Vec has been applied on a large movie review dataset to classify text so that the semantic associations between the terms stay conserved. A comparative study of the performances of some notable clustering algorithms is demonstrated concerning their application involving a variable number of features and classifier types as well as variable number of clusters.
Multilevel Quantum Sperm Whale Metaheuristic for Gray-Level Image Thresholding
Advances in Intelligent Systems and Computing, 2020
Image thresholding is a fundamental step in image segmentation. A clever selection of thresholds ... more Image thresholding is a fundamental step in image segmentation. A clever selection of thresholds is a vital step to achieve effective segmentation of images. In this article, we present a new quantum metaheuristic algorithm inspired by the behavior of sperm whales for optimal thresholding of gray-level images. The algorithm is built using many-valued quantum computing principles which offer greater computational advantages. Results are demonstrated on four test images with three threshold levels. The performance of the proposed algorithm has been compared with the qubit encoded quantum-inspired simulated annealing algorithm and the classical sperm whale algorithm with respect to the optimal fitness values and the computational time. Friedman test has been carried out with the competing algorithms to establish the supremacy of the proposed technique. Experimental results indicate the superiority of the proposed method in comparison with the competing methods.
Lecture Notes in Computer Science, 2019
A Quantum-Inspired Bidirectional Self-Organizing Neural Network (QIBDS Net) architecture operated... more A Quantum-Inspired Bidirectional Self-Organizing Neural Network (QIBDS Net) architecture operated by Quantum-Inspired Multi-level Sigmoidal (QIMUSIG) activation function suitable for fully automatic segmentation of T1-weighted contrast enhanced (T1-CE) MR images, is proposed in real time. The QIBDS Net architecture comprises input, intermediate and output layers of neurons represented as qubits and interconnected by second order neighborhood based topology. The inter-connections between the intermediate and output layers are effected by means of counter propagation of quantum states without any training or external supervision. Quantum observation is carried out at the end to obtain the segmented tumor from the superposition of quantum states. The proposed self-supervised network architecture has been tested on T1-CE MR images from publicly available data sets and is found to be very efficient while compared with other state of the art techniques.
An introductory illustration of medical image analysis
Advanced Machine Vision Paradigms for Medical Image Analysis, 2021
Abstract The medical imaging field has evolved into an enormous scientific discipline since the l... more Abstract The medical imaging field has evolved into an enormous scientific discipline since the last decade of the 19th century. The analysis of medical data obtained by current image modalities such as positron emission tomography, magnetic resonance imaging, computed tomography, and ultrasound comes to the aid of the fruitful diagnosis, appropriate planning, and assessment of therapy for patients’ treatment and much more. Medical image analysis is crucial to grip this huge amount of data and to investigate and present the appropriate information for any particular medical task. In this chapter, different aspects with regard to medical image analysis are exhaustively explored. In particular, issues and challenges in connection with this task are investigated and described. In addition, a brief summary of the contributory chapters is presented to trace the challenges and findings of each.
Chicago Crime Data Analysis Using PIG in Hadoop
2018 Fourth International Conference on Research in Computational Intelligence and Communication Networks (ICRCICN), 2018
The need for analyzing the different crime dataset has significantly increased over the past few ... more The need for analyzing the different crime dataset has significantly increased over the past few decades. It is necessary to detect the wide variety of crimes and the corresponding place of occurrences accurately. Government bodies throughout the world maintained Open Data initiatives, which is a large collection of heterogeneous dataset. This enables the government agencies to maintain the law and order of the society. The prime objective of this paper is to analyze the Chicago crime dataset to extract the significant crime information over the years. The proposed analysis is performed to bring out the crime information depending on some predetermined criteria, e.g. total crimes of different types, narcotic crime cases, an offense involving children, analysis of hourly theft cases and location identification where the major theft cases have occurred. The proposed analysis is performed in fully distributed Hadoop cluster. Moreover, we have extracted crucial minuscule statistics on t...
Relook into Sentiment Analysis performed on Indian Languages using Deep Learning
2018 Fourth International Conference on Research in Computational Intelligence and Communication Networks (ICRCICN), 2018
Sentiment analysis in various Indian languages has been gaining huge recognition recently for the... more Sentiment analysis in various Indian languages has been gaining huge recognition recently for the overall trend to take decisions based on reviews in almost all aspects of judgement making. With the sudden ascend of availability in the amount of data in the internet, extorting information from the opinions expressed by humans enhances in better understanding of their sentiments which is quite a challenging task. Recently, deep learning utilises its automatic feature learning techniques to outdo most of the state-of-the-art traditional methods of analysing sentiments. This paper aims to present the works that has already been done regarding analysis of sentiments in most available yet underrated Indian languages. Other than this, the probable problem domains are explored which can be solved in the area of multilingual sentiment analysis.
Introduction to Hybrid Metaheuristics
Series in Machine Perception and Artificial Intelligence, 2018
Segmentation of a video into its constituent shots is a fundamental task for indexing and analysi... more Segmentation of a video into its constituent shots is a fundamental task for indexing and analysis in content based video retrieval systems. In this paper, a novel approach is presented for accurately detecting the shot boundaries in real time video streams, without any a priori knowledge about the content or type of the video. The edges of objects in a video frame are detected using a spatio-temporal fuzzy hostility index. These edges are treated as features of the frame. The correlation between the features is computed for successive incoming frames of the video. The mean and standard deviation of the correlation values obtained are updated as new video frames are streamed in. This is done to dynamically set the threshold value using the three-sigma rule for detecting the shot boundary (abrupt transition). A look back mechanism forms an important part of the proposed algorithm to detect any missed hard cuts, especially during the start of the video. The proposed method is shown to...
ArXiv, 2020
There are numerous models of quantum neural networks that have been applied to variegated problem... more There are numerous models of quantum neural networks that have been applied to variegated problems such as image classification, pattern recognition etc.Quantum inspired algorithms have been relevant for quite awhile. More recently, in the NISQ era, hybrid quantum classical models have shown promising results. Multi-feature regression is common problem in classical machine learning. Hence we present a comparative analysis of continuous variable quantum neural networks (Variational circuits) and quantum backpropagating multi layer perceptron (QBMLP). We have chosen the contemporary problem of predicting rise in COVID-19 cases in India and USA. We provide a statistical comparison between two models , both of which perform better than the classical artificial neural networks.
Multi-verse Optimization Clustering Algorithm for Binarization of Handwritten Documents
Recent Trends in Signal and Image Processing, 2018
Binarization process of images of historical manuscripts is considered a challenge due to the dif... more Binarization process of images of historical manuscripts is considered a challenge due to the different types of noise that are related to the degraded manuscripts. This paper presents an automatic clustering algorithm for binarization of handwritten documents (HD) based on multi-verse optimization. The multi-verse algorithm is used to find cluster centers in HD where the number of clusters is predefined. The proposed approach is tested on the benchmarking dataset used in the Handwritten Document Image Binarization Contest (H-DIBCO 2014). The proposed approach is assessed through several performance measures. The experimental results achieved competitive outcomes compared to the well-known binarization methods such as Otsu and Sauvola.
IEEE Transactions on Neural Networks and Learning Systems, 2021
Classical self-supervised networks suffer from convergence problems and reduced segmentation accu... more Classical self-supervised networks suffer from convergence problems and reduced segmentation accuracy due to forceful termination. Qubits or bi-level quantum bits often describe quantum neural network models. In this article, a novel self-supervised shallow learning network model exploiting the sophisticated three-level qutrit-inspired quantum information system referred to as Quantum Fully Self-Supervised Neural Network (QFS-Net) is presented for automated segmentation of brain MR images. The QFS-Net model comprises a trinity of a layered structure of qutrits interconnected through parametric Hadamard gates using an 8-connected second-order neighborhood-based topology. The non-linear transformation of the qutrit states allows the underlying quantum neural network model to encode the quantum states, thereby enabling a faster self-organized counter-propagation of these states between the layers without supervision. The suggested QFS-Net model is tailored and extensively validated on Cancer Imaging Archive (TCIA) data set collected from Nature repository and also compared with state of the art supervised (U-Net and URes-Net architectures) and the self-supervised QIS-Net model. Results shed promising segmented outcome in detecting tumors in terms of dice similarity and accuracy with minimum human intervention and computational resources.
2020 IEEE Congress on Evolutionary Computation (CEC), 2020
Hyperspectral images contain a wide variety of information, varying from relatively large regions... more Hyperspectral images contain a wide variety of information, varying from relatively large regions to smaller manmade buildings, roads and others. Automatic clustering of various regions in such images is a tedious task. A multilevel quantum inspired fractional order ant colony optimization algorithm is proposed in this paper for automatic clustering of hyperspectral images. Application of fractional order pheromone updation technique in the proposed algorithm produces more accurate results. Moreover, the quantum inspired version of the algorithm produces results faster than its classical counterpart. A new band fusion technique, applying principal component analysis and adaptive subspace decomposition, is successfully proposed for the pre-processing of hyperspectral images. Score Function is used as the fitness function and K-Harmonic Means is used to determine the clusters. The proposed algorithm is implemented on the Xuzhou HYSPEX dataset and compared with classical Ant Colony Optimization and fractional order Ant Colony Optimization algorithms. Furthermore, the performance of each method is validated by peak signal-to-noise ratio which clearly indicates better segmentation in the proposed algorithm. The Kruskal-Wallis test is also conducted along with box plot, which establishes that the proposed algorithm performs better when compared with other algorithms.
Automatic Clustering of Hyperspectral Images Using Qutrit Based Particle Swarm Optimization
Intelligence Enabled Research, 2020
Hyperspectral Images (HSI) contain a lot of data channels. Due to their high dimensionality, it i... more Hyperspectral Images (HSI) contain a lot of data channels. Due to their high dimensionality, it is difficult to design systems which are able to find optimal number of clusters to segment them. A qutrit based particle swarm optimization (PSO) for automatic clustering of hyperspectral images is introduced in this paper. A Band Fusion Technique is implemented by improving the Improved Subspace Decomposition Algorithm using SF Index as the fitness function. A new method for maintaining the superposition state of the qutrits is also successfully designed. A comparison with classical PSO is performed to clearly establish the supremacy of the proposed technique with respect to Peak signal-to-noise ratio (PSNR), Jaccard Index, Sorensen-Dice Similarity Index and the computational time. Finally, the unpaired two-tailed t test is conducted between the proposed technique and classical PSO, which renders better results for proposed qutrit based technique. The experiments are carried out on the Salinas Dataset. The proposed technique yields better results in all the tests conducted in comparison to the classical PSO.
Uploads
Papers by Siddhartha Bhattacharyya