Exploring Word2Vec for query expansion for ad-hoc retrieval.lg...Expand

Project 1 : Language Detection for Devanagari Script It is a NLP tool to detect language given a sentence, The language it can detect is of the Devanagari Script : Nepali, Marathi, Hindi. These languages are very close when comes to the word usage and semantic patterns. The key challenge we faced is how to do the classification at word level without looking at utf-8 code. We got an overall accuracy of 97% on the test set. We have used weighted N-Gram with Laplace smoothing and Naive Bayes classifier. Project 2: English Pronoun Anaphora resolution using word embedding and translation model Pronoun Anaphora resolution is one of the challenging task in natural language processing. In the above project, we have addressed the anaphora resolution in English language for detecting noun pronoun referencing. We have implemented a word embedding model with a language translator for the same. The baseline results we got, were satisfactory by there is enough scope for improvement. Hence it is being worked upon by applying advanced models and transfer learning Show lesslg...Expand

We tried to analyze how facebook usage patterns varies with personality. How people get affected by and react to different scenarios of facebook interaction.lg...Expand

This project presents a short term load forecasting technique for a microgrid based upon smart grid architecture. An efficient demand management and demand response requires dissemination of the historical pattern of consumption information to the consumer level. This information will enable the customer to appropriately tune their load through implementation of energy management systems. An efficient and economic central load forecasting system which can predict the several hours ahead load for the microgrid as well as for each individual customer is proposed in this paper. Implementation of this method will be able to reduce the cost of an individual forecasting system for each consumer and the microgrid. Show lesslg...Expand

As a part of my dissertation, I am working with Natural Language Processing and Deep Neural Networks. I have exploted Convolutional Neural Nets, Recurrent Neural Nets using Theano. Multi Layer Perceptron, using Theano and in general Python Self Organizing Map PCA Analyzer etc.lg...Expand