Nilanshi Chauhan

• Actively engaged in investigating, reviewing and identifying insights for non intrusive load monitoring (NILM) based business model for electrical utilities and customers.
• Vision is to pioneer platform that crafts personalized plans, uniquely tailored to optimize energy consumption, reduce expenses, and preemptively identify faulty devices for our valued customers.

Project: Machine Learning Application in Wireless Sensor Networks.

• Proposed and devised a fully distributed game-theory based reinforcement learning technique to restore coverage loss in shortest possible time span and minimum coverage overlap.
• Optimized sensor spatial density of randomly deployed nodes for three tessellations using mathematical analysis.

Project: Application of Deep Learning to Predict and Counter Forest Fires

• Researched environmental conditions (usually a combination of dryness, heat wave and high winds) causing wildfires. Proposed a deep reinforcement learning based technique for sensor nodes to predict forest fires.
• Proposed area prioritization algorithm based on gradient descent to assess and rank areas based on their vulnerability to damage, population density, presence of animals, real estate, and fire risk.

Project: Strategic Implementation of Machine Learning for Opinion Mining in Online Product Reviews.

• Generated a feature-based organized summary for online expressed opinions, depending on frequency of occurrence of potential features.
• Implemented pruning methods and Apriori algorithm based supervised learning to obtain final feature set for sentiment analysis and opinion summary generation using Natural Language Processing Toolkit (NLTK).
• Achieved average accuracy of 78.68% for feature extraction, 83.45% for opinion word extraction and 94.58% for opinion word orientation determination.

• Supervised and conducted laboratory sessions for more than 60 undergraduate students.
• Presented research findings at national and international conferences, raising profile of institution.
• Studied and implemented latest information on specific subjects and learning strategies.