- Mumbai, Maharashtra, India
KARTHIKA RENUKA
PSG college of Technology, Information Technology, Faculty Member
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Emotions are complex phenomena that play significant roles in the quality of human life. Emotion plays a major role in motivation, perception, cognition, creativity, attention, learning and decision-making. A major problem in... more
Emotions are complex phenomena that play significant roles in the quality of human life. Emotion plays a major role in motivation, perception, cognition, creativity, attention, learning and decision-making. A major problem in understanding emotion is the assessment of the definition of emotions. According to the WHO, every year, almost one million people die from suicide. Suicide is a leading cause of death among teenagers and adults. Existing techniques uses simple keyword search method to find emotional content in blog data and identify bloggers at risk of suicide. However, Deep sentiment analysis in suicide notes has not yet been explored much with computational approaches using advanced Machine Learning and Natural Language Processing techniques. The main contribution of the proposed work employs Electroencephalography (EEG) based psychological states for initializing the parameter weights of the neural network, which is crucial to train an accurate model while avoiding the need...
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Multimodal analysis focuses on the internal and external manifestations of cancer cells to provide physicians, oncologists and surgeons with timely information on personalized diagnosis and treatment for patients. Decision fusion in... more
Multimodal analysis focuses on the internal and external manifestations of cancer cells to provide physicians, oncologists and surgeons with timely information on personalized diagnosis and treatment for patients. Decision fusion in multimodal analysis reduces manual intervention, and improves classification accuracy facilitating doctors to make quick decisions. Genetic characteristics extracted on biopsies do not, however, provide details on adjacent cells. Images can only provide external observable details of cancer cells. While mammograms can detect breast cancer, region wise details can be obtained from ultrasound images. Hence, different types of imaging techniques are used. Features are extracted using the SelectKbest method in the Wisconsin Breast Cancer, Clinical and gene expression datasets. The features are extracted using Gray Level Co-occurrence Matrix from Histology, Mammogram and Sonogram images. For image datasets, the Convolution Neural Network (CNN) is used as a cl...
Research Interests:
Research Interests:
Emotional AI is the next era of AI to play a major role in various fields such as entertainment, health care, self-paced online education, etc., considering clues from multiple sources. In this work, we propose a multimodal emotion... more
Emotional AI is the next era of AI to play a major role in various fields such as entertainment, health care, self-paced online education, etc., considering clues from multiple sources. In this work, we propose a multimodal emotion recognition system extracting information from speech, motion capture, and text data. The main aim of this research is to improve the unimodal architectures to outperform the state-of-the-arts and combine them together to build a robust multi-modal fusion architecture. We developed 1D and 2D CNN-LSTM time-distributed models for speech, a hybrid CNN-LSTM model for motion capture data, and a BERT-based model for text data to achieve state-of-the-art results, and attempted both concatenation-based decision-level fusion and Deep CCA-based feature-level fusion schemes. The proposed speech and mocap models achieve emotion recognition accuracies of 65.08% and 67.51%, respectively, and the BERT-based text model achieves an accuracy of 72.60% . The decision-level ...
Research Interests:
Speech recognition has now become ubiquitous and plays an inevitable role in almost all sectors. Numerous works have been proposed on speech recognition; however, more accurate transcriptions are not possible. Exploration of various... more
Speech recognition has now become ubiquitous and plays an inevitable role in almost all sectors. Numerous works have been proposed on speech recognition; however, more accurate transcriptions are not possible. Exploration of various studies related to spell correction implies that several kinds of research have been carried out in this field but still it is a very challenging problem. This led to the need for a new spell corrector framework capable of leveraging the performance of the automatic speech recognition (ASR) system. The proposed work unveils state-of-the-art Bidirectional Encoder Representations from Transformers (BERT) based spell correction module developed on top of the deep recurrent neural network (RNN) based ASR system. The impact of BERT-based spell correction on the ASR system is evaluated on three different accent datasets in the perspective of word error rate (WER), character error rate (CER), and Bilingual evaluation understudy (BLEU) score. The experimental re...
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Cloud computing is the cutting edge technology in the information field to provide services to the users over the internet through web–based tools and applications. One of the major aspects of cloud computing is load balancing. Challenges... more
Cloud computing is the cutting edge technology in the information field to provide services to the users over the internet through web–based tools and applications. One of the major aspects of cloud computing is load balancing. Challenges like Quality of service (QoS) metrics and resource utilization can be improved by balancing the load in cloud environment. Specific scheduling criteria can be applied using load balancing for users prioritization. This paper surveys different load balancing algorithms. The approaches that are existing are discussed and analyzed to provide fair load balancing and also a comparative analysis was presented for the performance of the existing different load balancing schemes.
Research Interests:
By automatically translating Indian sign language into English speech, a portable multimedia Indian sign language translation program can help the deaf and/or speaker connect with hearing people. It could act as a translator for those... more
By automatically translating Indian sign language into English speech, a portable multimedia Indian sign language translation program can help the deaf and/or speaker connect with hearing people. It could act as a translator for those that do not understand sign language, eliminating the need for a mediator and allowing communication to take place in the speaker's native language. As a result, Deaf-Dumb people are denied regular educational opportunities. Uneducated Deaf-Dumb people have a difficult time communicating with members of their culture. We provide an incorporated Android application to help ignorant Deaf-Dumb people fit into society and connect with others. The newly launched program includes a straight forward keyboard translator that really can convert any term from Indian sign language to English. The proposed system is an interactive application program for mobile phones created with application software. The mobile phone is used to photograph Indian sign languag...
Research Interests:
Research Interests:
A precise seizure detection system allows epileptic patients to receive early warnings before a seizure occurs. It is critical for people who are drug-resistant. To find the very minimal time before seizure onset, traditional seizure... more
A precise seizure detection system allows epileptic patients to receive early warnings before a seizure occurs. It is critical for people who are drug-resistant. To find the very minimal time before seizure onset, traditional seizure prediction techniques rely on variables collected from electroencephalography (EEG) recordings and classification algorithms. Such methods cannot achieve high-accuracy prediction due to the information loss of hand-crafted features and the limited classification capabilities of regression and other algorithms. Kernels are employed in the early and late stages of the CNN RNN architecture with VGG 16 in the convolution and max-pooling layers, respectively. The suggested hybrid model is tested using the CHB-MIT scalp EEG datasets. The total sensitivity, false prediction rate, and area under the receiver operating characteristic have all yielded positive results.