Digital CEO
Introducing Digital CEO, your voice-enabled solution for operational excellence. With Digital CEO, leaders can effortlessly pose questions and promptly receive curated insights across various substrates and solutions. This pioneering project revolutionizes decision-making and operational efficiency by providing leaders with swift and precise access to relevant information. Leveraging embedding models, VectorDBS, Large Language Models (LLMs), and the RAG (Retrieval-Augmented Generation) framework, Digital CEO streamlines processes and empowers data-driven decision-making like never before. Experience the future of leadership with Digital CEO, where insights are readily accessible through the power of voice technology.
AI-Tutor
Discover AI-Tutor, your personal learning companion for educational videos. With real-time assistance and personalized feedback, AI-Tutor transforms passive viewing into an engaging, interactive experience. Whether you're a student seeking clarity or an educator enhancing your lessons, AI-Tutor makes learning dynamic and effective.
Product Knowledge Graphs
Created knowledge graphs using NLP and graph-based techniques to extract valuable insights from product reviews and comments. The Product Knowledge Graphs project significantly increased business profits by leveraging insights extracted from customer con-versational data. By modeling the data into graphs and analyzing them, the organization gained valuable information about product preferences, user sentiments, and market trends, leading to informed business decisions.
Project: Key Phrase Extraction
Developed algorithms using graph properties such as centrality, degree, and PageRank to extract key phrases from conversational data. The Key Phrase Extraction project improved data understanding by automatically extracting key phrases, which made information easier to manage, classify, and retrieve. This enhanced data organization and enabled efficient analysis, leading to improved decision-making and targeted actions.
Performance Evaluation of Sentiment Analysis on Text and Emoji Data using EndtoEnd, Transfer Learning, Distributed and Ex-plainable AI Models
Conducted sentiment analysis on text and emoji datasets, evaluating various models and techniques.The Performance Evaluation project provided valuable insights into sentiment analysis techniques on different types of data. By evaluating end-to-end, transfer learning, distributed, and explainable AI models, the organization gained a deeper understanding of sentiment analysis and its applications, enabling more accurate and reliable sentiment analysis in various contexts.
Intelligent Computing for Skill-set Analytics in a Big Data Framework
Applied intelligent computing techniques to analyze skill-set data in a big data framework. The Intelligent Computing project facilitated skill-set analytics by utilizing big data frameworks and graph modeling. By extracting intelligence from a large volume of resumes, the organization gained insights into skill trends, identified skill gaps, and optimized talent acquisition strategies.
Cryptographic Algorithm Identification Using Deep Learning Techniques
Developed deep learning models to identify cryptographic algorithms from cipher texts. The Cryptographic Algorithm Identification project enhanced security measures by accurately identifying encryption algorithms using deep convolutional neural networks. This enabled the organization to strengthen its cryptographic protocols and protect sensitive information effectively.
Pothole Detection Using Deep Convolutional Neural Network
Explored the possibility of developing a Deep Learning model which can detect potholes on roads in real time with maximum accuracy and minimum inference delay.
Novel Graph Based Anomaly Detection Using Background Knowledge
In this project, I proposed a groundbreaking approach for graph-based anomaly detection by incorporating background knowledge into the evaluation metrics. By biasing the substructure discovery process towards identifying anomalous patterns, we achieved comparable accuracy and search space to existing methods. Our novel approach to graph-based anomaly detection opens up new possibilities in identifying and mitigating potential risks within complex systems. By leveraging background knowledge, we were able to enhance the accuracy and effectiveness of anomaly detection algorithms, providing valuable insights for improved decision-making and system security.
Discovering Suspicious Patterns Using a Graph Based Approach
In this project, our goal was to identify fraudulent employee activities within Kasios, a furniture manufacturing company, by employing a graph-based approach. By analyzing the data and visualizing suspicious patterns using the enterprise graph database Neo4j, we aimed to uncover fraudulent activities and enhance security measures. By utilizing a graph-based approach, we contributed to the field of fraud detection and prevention, particularly in the context of employee activities. Our findings and methodologies provided valuable insights into identifying suspicious patterns and strengthening organizational security measures.
Compressing Graph Data by Leveraging Domain Independent Knowledge
In this project, we focused on graph compression techniques to improve the visualization and understanding of high-level graph structures. We introduced CRADLE (CompRessing grAph data with Domain independent knowLEdge), a novel method that employed knowledge rules, specifically netting, to report the number of external networks for each substructure instance. Our research on graph compression techniques contributes to the effective analysis and interpretation of complex graph data. By leveraging domain-independent knowledge, we provided a valuable tool for graph visualization and understanding, enabling researchers and practitioners to gain meaningful insights from large-scale graph datasets.
Mining Interesting Substructures using Graph Properties as Background KnowledgeÂ
In this project, we introduced an innovative approach to mine interesting substructures by incorporating graph properties as background knowledge in the evaluation metrics. By considering factors such as density, degree, and other relevant properties, we were able to bias the substructure discovery pro-cess towards identifying patterns of interest. The outcomes of this research were published in the Intelligent Data Analysis Journal, a prestigious SCI indexed journal. Our approach to mining interesting substructures using graph properties provided a valuable tool for uncovering meaningful patterns and insights from complex graph datasets. By leveraging background knowledge, we enhanced the effectiveness and efficiency of substructure discovery, enabling researchers and practitioners to gain deeper understanding and make informed decisions based on graph analytics.
A Graph Based Approach for IP Network Analysis
In this project, we developed a graph-based approach for analyzing network data and identifying structural patterns. By leveraging the complete profile of the computer infrastructure, our approach enabled us to detect potential vulnerabilities and alert network administrators in preventing exploitation. Our graph-based approach for IP network analysis contributed to the field of network security by providing an effective tool for identifying structural patterns and potential vulnerabilities. By alerting network administrators in a timely manner, we helped mitigate risks and enhance the security of computer infrastructures.
Proposals of Graph Based Ciphers using Matrix Algebra and Minimum Spanning Trees
In this project, we proposed innovative methods for designing effective substitution ciphers by utilizing paths, weights, and matrix algebra between pairs of graph vertices. By leveraging the advantage of matrix algebra and minimum spanning trees, we developed novel graph-based ciphers. Our proposals of graph-based ciphers using matrix algebra and minimum spanning trees offered new perspectives and techniques for secure encryption. By combining graph theory and cryptography, we contributed to the advancement of cryptographic algorithms, enhancing data security in various domains.
Context Based Fake News Detection using Graph Based Approach: A COVID19 Usecase
In this project, we addressed the challenge of detecting fake news articles by employing a contextual graph-based approach. By converting news articles into contextual graphs using Natural Language Processing techniques, we aimed to identify and combat misinformation during the COVID-19 pandemic. Our research on context-based fake news detection using a graph-based approach provided a valuable tool for combating misinformation and promoting information integrity. By leveraging graph representations and NLP techniques, we contributed to the development of robust methods to identify and mitigate the spread of fake news, particularly during critical events such as the COVID-19 pandemic.