Purnendu Bikash Acharjee
@lavasa.christuniversity.in
Associate Professor, Computer Science
CHRIST university
Dr. Purnendu Bikash Acharjee has 16+ years of academic and research experience. Currently, he is working as an Associate Professor in the School of Business and Management, CHRIST [Deemed to be University], Pune. Prior to that, he was an Associate Professor, at the School of Computing, RIMT University, Punjab, and an Associate Professor, Associate Dean, at the School of Computing Sciences, Jorhat, Assam. Dr. Acharjee obtained M.Sc. (Computer Science), MTech. (IT), and Ph.D. (Faculty of Technology) from Gauhati University, Assam. Dr. Acharjee has 37 papers [14 SCOPUS indexed], 4 [SCI indexed], 3 book chapters [refereed international and national journals], 1 book and granted 6 International Patents, and 7 National Patents got published. Participated in 22 AICTE/UGC/Industry sponsored (two weeks duration) short-term courses and FDPs, Paper Reviewer in 4 International Conferences, and Technical Program Committee member in 7 International Conferences including IEEE, Springer, etc. He also
EDUCATION
MSc in Computer Science, MTech in Information Technology, PhD under Faculty of Technology
RESEARCH INTERESTS
Speech Processing, Speech Recognition, Big Data Analytics
Scopus Publications
Scholar Citations
Scholar h-index
Scholar i10-index
Scopus Publications
Sippee Bharadwaj and Purnendu Bikash Acharjee
Springer Science and Business Media LLC
Tushar Dhar Shukla, Purnendu Bikash Acharjee, Chethan C, Thulasimani T, M Sindhu, and Sanjiv Sharma
IEEE
Undergraduate and postgraduate students from eight different departments at a UK institution participated in organized conversations about the impact of teachers' research activities on their education. In both samples, positive responses greatly outnumbered negative ones. There was an increase in positive feedback on professors' research when the overall quantity and quality of research in a specific field (as measured by Research Assessment Exercise [RAE] ratings) improved. Undergraduate samples with higher RAE scores were more likely to have negative feedback on research than graduate student samples. Both graduate and undergraduate students agreed that lecturers' research increased the instructor's credibility, relevance, and knowledge, as well as piqued and maintained their own interest, engagement, and drive. Data processing, feature selection, and model training are the first steps in the proposed approach. The data are changed from their raw form into a form suitable for academic use during the data pre-processing phase. They are employing Information Gain and Symmetric Uncertainty for feature selection. Following the feature selection process, the models are trained using BiLSTM-CNN. Both the BiLSTM and the CNN methods are inferior to the proposed method.
Purnendu Bikash Acharjee, Alkawati Annappa Magadum, Murari Thejovathi, Renu Jain, K Umarani, and Neerav Nishant
IEEE
Sentiment, volumetric, and social network analyses, as well as other methods, are examined for their ability to predict key outcomes using data collected from social media. Different points of view are essential for making significant discoveries. Social media have be en used by individuals all over the world to communicate and share ideas for decades. Sentiment analysis, often known as opinion mining, is a technique used to glean insights about how the public feels and thinks. By gauging how people feel about a candidate on social media, they can utilize sentiment analysis to predict who will win an upcoming election. There are three main steps in the proposed approach, and they are preprocessing, feature extraction, and model training. Negation handling often requires preprocessing. Natural Language Processing makes use of feature extraction. Following the feature selection process, the models are trained using BiCNN-RNN. The proposed method is superior to the widely used BiCNN and RNN methods.
BMG Prasad, Purnendu Bikash Acharjee, Sivaprasad Guntakala, Deepak Sharma, N Divya, and Harshal Patil
IEEE
The malignant bone tumor osteosarcoma. Any bone is at risk, but lengthy bones like the limbs are more vulnerable. Although the precise cause of this malignant growth is uncertain, experts concur that it is caused by changes to deoxyribonucleic acid (DNA) inside the bones. This can cause the breakdown of good tissue and the growth of aberrant, pathological bone. Osteosarcoma has a 76% cure rate if detected early and treated before it spreads to other parts of the body. An X-ray is the primary tool for detecting bone tumors. Bone X-rays and other imaging tests can help detect osteosarcoma. A biopsy should be performed for an accurate diagnosis. This is a time-consuming and tedious task that might be greatly reduced with the help of appropriate tools. Data preprocessing, segmentation, feature extraction, and model training are the four main pillars of the proposed approach. Unwanted noises can be filtered out with some preprocessing. Low-spatial-frequency and high-spatial-frequency components are separated using segmentation. The proposed approach employed Tumor Border Clarity, Joint Distance, Tumor Texture, and other features for feature extraction. Let's move on to A-Residual CNN model training. The success percentage of the proposed approach was 96.39 percent.
Purnendu Bikash Acharjee, Manish Kumar, Gopal Krishna, Kamalakar Raminenei, Read Khalid Ibrahim, and Malik Bader Alazzam
IEEE
Cyber-attacks have become a growing concern for governments, organizations, and individuals worldwide. In this paper, we explore the use of blockchain technology to secure international law against cyber-attacks. We discuss the advantages of blockchain technology in providing secure and transparent data storage and transmission, and how it can enhance the security of international law. We also review the current state of international law regarding cyber-attacks and the need for a robust and effective legal framework to address cyber threats. The study proposes a blockchain-based approach to secure international law against cyber-attacks. We examine the potential of blockchain technology in providing a decentralized and tamper-proof database that can record and track the implementation of international laws related to cyber-attacks. We also discuss how smart contracts can be utilized to automate compliance with international laws and regulations related to cybersecurity. The study also discusses the challenges and limitations of using blockchain technology to secure international law against cyber-attacks. These include the need for interoperability between different blockchain networks, the high energy consumption of blockchain technology, and the need for international cooperation in implementing and enforcing international laws related to cybersecurity. Overall, this study provides a comprehensive overview of the potential of blockchain technology in securing international law against cyber-attacks. It highlights the need for a robust legal framework to address cyber threats and emphasizes the importance of international cooperation in implementing and enforcing international laws related to cybersecurity.
R. Sharmila, R Kamalitta, Moorthy, Devesh Pratap Singh, Amit Chauhan, and Purnendu Bikash Acharjee
IEEE
Large losses in output, money, and quality/quantity of agricultural goods are incurred due to plant diseases. Seventy percent of India’s GDP is tied to the agricultural sector, thus protecting plants from diseases is crucial. For this reason, it is important to keep an eye on plants from the moment they sprout. The usual approach for this omission is naked eye inspection, which is more time-consuming, costly, and requires significant skill. Thus, automating the method for detecting diseases is necessary to speed up this process. It is imperative that image processing methods be used in the creation of the illness detection system. Disease detection involves a number of processes, including Weighted Mask R-CNN, GLCM feature extraction, Multi-thresholding image pre-processing, and K means image segmentation classification. The weighted Mask R-CNN outperforms the standard RNN, the Mask R-CNN, and the CNN in terms of accuracy and recall in analytical trials by a significant margin.
S. Murugesan, Neelam Sharma, N. Jayalakshmi, Kumud Pant, Amit Chauhan, and Purnendu Bikash Acharjee
IEEE
A major focus of current research in the field of image processing is the application of such methods to the field of medical imaging. While dealing with biological issues like fractures, canoers, ulcers, etc., image processing facilitated pinpointing the precise cause and tailoring a remedy. In the field of tumor identification, medical imaging has set a new standard by overcoming a number of challenges. Medical imaging is the practice of generating images of the human body for diagnostic or exploratory purposes. Because of its high image quality, MRI is the method of choice for detecting tumors. This research study proposes the integration of RLM to detect tumors and presents an automatic bone cancer detection system to assist oncologists in making early diagnosis of bone malignancies, which in turn allows patients to receive treatment as soon as possible. This research work also proposes to detect bone tumors by using a combination of the RELM based M3 filtering, Canny Edge segmentation, and the Enhanced Harris corner approach. When compared to other models like CNN, ELM, and RNN, the suggested technique achieves an accuracy of around 97.55%.
Yousef Methkal Abd Algani, S. Vidhya, Bhupaesh Ghai, Purnendu Bikash Acharjee, Mathur Nadarajan Kathiravan, and Vijay Kumar Dwivedi
IEEE
Brain illnesses are notoriously challenging because of their fragility, surgical complexity, and high treatment costs. Contrarily, it is not obligatory to carry out the operation, as the outcomes of the procedure may fall short of expectations. Adult-onset Alzheimer's disease, which causes memory loss and losing information to varied degrees, is one of the most common brain diseases. This will vary from person to person based on their current health situation. This highlights the need of using CT brain scans to classify the extent of memory loss and determine the patient's risk for Alzheimer's disease. The four main goals of Alzheimer's disease detection are preprocessing the data, extracting features, selecting features, and training the model with GP-ELM-RNN. The Replicator Neural Network has been utilized earlier for AD detection, however this study offers an improved version of the network, modified with ELM learning and the Garson algorithm. From this study, it is deduced that the proposed method is not only efficient, but also quite precise. In this research, GP-ELM-RNN network is built to four groups of images representing different stages of Alzheimer's disease: very mildly demented, mildly demented, averagely demented, and non-demented. The class of very mildly demented patients was found to have the highest accuracy (99.1%) and specificity (0.984%). As compared to the ELM and RNN models, this technique achieves superior accuracy (around 99.23%).
Sonali Vyas, A. Narasima Venkatesh, Purnendu Bikash Acharjee, Satish G. Jangali, Charanjeet Singh, and Zarrarahmed Z. Khan
Springer Nature Singapore
Somi Kolita and Purnendu Bikash Acharjee
Springer Nature Singapore
Sunita Rani, Purnendu Bikash Acharjee, Suresh Kaswan, Vijay Anant Athavale, M. Udhayamoorthi, and Kumud Pant
IEEE
Over the years, technology played an integral role in enhancing eventual data management and also provided effective user satisfaction. Facial biometrics happen to be one of the most effective forms of face spoofing detection which significantly enhance data management to the next level. It is to be seen that photos and videos are the fundamental supplant for this system which essentially complement the data collection, storing and evaluation process. Using this particular technology user are able to utilise their devices for data management by storing and recording their personal photos into the devices. Now, the device with face spoofing detection would essentially identify users based on their historical data in the device database. This also allows user autonomy at the highest level and that is considered as one of the major reasons behind enhanced demand for this system. The research gas used positivism philosophy, deductive approach and descriptive research design. In addition, primary data collection and quantitative data analysis are also used in this research. A total of 50 participants were considered for the survey which is chosen as a primary data collection method. Based on the data analysis and discussion the crucial role of machine learning can be understood in terms of facilitating face spoofing detection.
G. S. Raghavendra, S. Shyni Carmel Mary, Purnendu Bikash Acharjee, V. L Varun, Syed Nisar Hussain Bukhari, Chiranjit Dutta, and Issah Abubakari Samori
Hindawi Limited
In the era of digital technology, where innovation and digitization are transforming the business functions, the field of customer relationship management has witnessed sea change in the recent decade. The application of artificial intelligence in food processing has enabled enhancing the availability of food in an effective manner for all the individuals. It has been regarded that the application of labor force tends to play a crucial aspect for the overall execution of things in the different domains related to food manufacturing and processing, which are related to the enhanced involvement of individuals in the processing of food and related products, the industry could not able to meet the growing demand from the customers. So, in order to overcome these critical issues, it is noted that the application of technology such as automation and artificial intelligence is implemented for enhanced processing and enable delivering quality products to the customers at lower cost. The impact of AI in the current business world is becoming more indispensable as companies have started to unleash its potential. The role of AI in food processing is fast changing the manner in which the customer queries are addressed, enabling analysing the needs and requirements, and focus on creating improved packaging, high quality, and better shelf life. This empirical investigation is focusing on analysing the critical factors related to AI in influencing the food processing industry. The researchers intend to apply quantitative analysis using IBM SPSS package, and the results are stated in detail based on the analysis.
Shahanawaj Ahamad, Priti Gupta, Purnendu Bikash Acharjee, K. Padma Kiran, Zarrarahmed Khan, and Mohammed Faez Hasan
Elsevier BV
Atul Kumar Dwivedi, , Deepali Virmani, Anusuya Ramasamy, Purnendu Bikash Acharjee, Mohit Tiwari, , , , and
Elsevier BV
Speech is considered as the fundamental aspect of communication between human being, speech recognition is stated as the overall process to convert the sound into corresponding text based on a specific language. The implementation of speech recognition has supported individuals, business and others in order to possess better communication and interaction so as to realise its objectives. This been regarded as the process of collating text message or in some form of meaning based on the input received from voice of another individual. The speech analytics is stated as the key part in the speech recognition as it converts the individual voice into digital form so as to store them and transmit it as and when required using computing equipments.The speech synthesis is considered as the reversal of speech recognition as they convert the data from the digitised format into voice which supports the users to listen quickly and easily.The application of speech recognition in organisation is confined in building more interactive virtual assistants, supports the customers in addressing their queries and offer solutions at quick span of time, furthermore organisations can use speech recognition to identify the individuals so that they can access classified information or reset their password etc. The enhanced development in the technology domain has deepened the importance of artificial intelligence in different areas of work and life, The implementation of AI in speech recognition supports the business and individuals in apprehending better services to the stakeholders and perform the task in an efficient manner. Hence, this study is focused in analysing the key determinates of using AI in speech recognition for effective multifunctional Machine learning platform using regression analysis.
M. Kiruthiga Devi, Veena Prasad Vemuri, Mahalakshmi Arumugam, S. K. UmaMaheswaran, Purnendu Bikash Acharjee, Rupali Singh, and Karthikeyan Kaliyaperumal
Hindawi Limited
In the current information and technology era, business enterprises are focusing in performing the process effectively by reducing the waiting time in completing the work, reduce latency and deploy the resources effectively so as to service the patient, medical practitioners, societies, and other stakeholders in an efficient manner. Hence, several organisations are using the emerging technologies so as to obtain high performance and enhance competitive edge. The advancement in machine learning, deep learning, business analytics, etc. enables the health care industry to identify the patterns based on the data collected and create a pivotal position and enhance revenues and profits in a sustainable manner. Machine learning models are considered as computational algorithms which will enable in collected the data, analyze them, and provide the necessary reports to the experts and management in order to make informed decision making. The application of advanced machine learning enables the organisation to process the image effectively, recognize the voice and enable in servicing the customers, process the available data, and identify the patterns so as to make informed decision making. The basic purpose of the study is to analyze the overall implementation of advanced machine learning approaches towards health care services for providing enhanced services, better patient engagement, and support in creating better life for them, the researchers intend to collect the closed-ended questionnaire from employees in different medical centers covering: apprehend the nature of designing and implementation of machine learning approaches in the organisation and understand the effectiveness of these tools in enhancing the sustainable growth and development of the organisation.
Sippee Bharadwaj and Purnendu Bikash Acharjee
IEEE
the present study is an attempt to explore the real-time applications of suprasegmental (prosodic) features in psychological intervention. The main purpose of this research is to explore different emerging speech analyzing tool for assessing stress. Earlier research work on medical science has shown that speech features can be used as an indicator of stress severity including anxiety disorders, depression, chronic pain, heart disease, asthma, autoimmune diseases, and neurodegenerative disorders. The research work is carried out on simulated controlled Assamese speech rather than natural conversation. It is investigated and observed from the speech signal that the prosodic features are the best indicators for accessing stress in therapeutic intervention to understand the mental state of a speaker. The variation in the supra-segmental speech features pitch, duration, and energy are computed from the emotional utterance to study the psychological condition of the speaker. These parameters are further extracted separately from different levels such as entire utterances, words, and syllables. This analytical non-invasive study is a small contribution in the area of speech processing in medical science.
Sippee Bharadwaj and Purnendu Bikash Acharjee
IEEE
This research work has attempted to study the speech prosody for designing a speech synthesizer. Speech prosody carries all relevant information about the utterances. Here, it has dealt with the suprasegmental features of the speech utterance. The propsoed research work is experimentally studying the interaction and behavior of segmental and suprasegmental features by concerning the North-East Indian Language Assamese. For efficiently designing an emotional speech recognition system or a speech synthesizer, a proper clarity on understanding the tone of a voice, boundaries of the prosodic, prominent portion of utterance concerning specific words, sentences are required. This paper also discusses the variations of prosodic features for ‘anger’ ‘normal’ ‘sad’ ‘surprise’ ‘disgust’ for different accent types. The degree of emotion considered here are head-high, mid-high, and flat with the help of some selected speech items for female and male taken from our Assamese speech database. This research work has analyzed and discussed the prosodic behavior of utterances from the mean speech rate and fundamental frequency based on the graphs and tables derived from the proposed experiment.
Palash Dutta, Laba Kr. Thakuria, Akalpita Das, Purnendu Acharjee, and Pranhari Talukdar
IEEE
The Intonation modeling has an important role of speech recognition and speech synthesis. Through this paper we have summarized the basic concepts of prosody, which deals with the outlines of prosody modeling approaches to the generation of prosodic information with the help of three different prosody models: Fujisaki, ToBI and Tilt. The main aim and significance of this research paper of Assamese prosody modeling and its implementation to Text-To-Speech systems are evaluated.
Akalpita Das, Laba Kr. Thakuria, Purnendu Acharjee, and P.H. Thakdar
IEEE
A model is proposed to developed a Indigenous language (Galo) sentence's pitch-contour with sentence-wide optimization, called the sentence pitch-contour using HMM(Hidden Markov Model) & VQ (vector quantization). To develop a sentence pitch-contour (SPC-HMM), each training sentence are normalized for the pitch-contours of the syllables. Our model is effective for pitch height normalization. After the process of normalization completed, the pitch-contour of each training syllable is vector quantized(VQ). The lexical tones of adjacent syllables and the quantization code are combined both to define the observation symbol sequences for HMM training. Using a dynamic-programming based algorithm in the synthesis phase, for given a sentence and related text-analysis information, the probable observation sequence is generated by finding the sentence wide largest probability path. We have also conducted practical perception tests and is observed that the speech synthesized using the sentence pitch-contour (SPC-HHM) is better than uttered by an ordinary speaker. To get the naturalness of the synthesized speech pitch counter has a great role.
Laba Kr. Thakuria, Purnendu Acharjee, Akalpita Das, and P.H. Thakdar
IEEE
This research paper represents a hybrid technique to improve the quality of the rule-based approach to generate prosody for Bodo speech synthesis by integrating prosody parametric manipulation with template parametric manipulation to increase the intonation variability of the synthesized output. Prosody means the rhythmic and intonation aspects of a spoken language. Prosody is also the combination of pitch, intensity and duration. In our normal speech, prosodic characters can be evaluated to express different emotions. Generally prosodic characters of neutral synthesized speech are evaluated to express the four basic emotions. We have considered happiness, anger, sadness and fear. Through this concept we are going to represent a methodology to express and evaluate the effect of the output to produce the accurate prosody of Bodo speech to confirm the perception tests. The correlation between acoustic features and emotion in Bodo speech are ascertained by the analysis of the database of Bodo speech.
Purnendu Bikash Acharjee, Jyotismita Talukdar, Akalpita Das, and P. H. Talukdar
IEEE
In this paper we are reporting the dependency of dialect variation while designing the G2P (Grapheme to Phoneme) rules for the Bodo language. The present study is based on the Bodo language, a language originated from the Sini-Tibetan family of languages. Language problem is becoming a major issue in all the ethnic disturbances in this part of India. Bodo language has a very little exposure in the greater national context. They are changing their scripts very often. Nearly 30 lakhs Bodo speaking people are spreading out the entire north-eastern region of India. So when they raised their demand for separate language identity from other ethnic groups of this region few factors becomes very important to establish ther identity in language domain. Because the ethnic groups Rabha, Kachari, Dimasa etc belong to same sino-Tibetan family of language and their languages are very close to each other. The present study primarly deals the issues of dialect variation within the same Bodo speaking people so that they can be isolated from nearest ethnic group in their linguistic domain. Finally an attempt has also been made to design the G2P rule for the Bodo language.