Machine Learning India bio photo

Machine Learning India

Fostering science and machine learning in India

Email Mailing List Twitter Github

Are you a student, professor, CEO or Maschinenmensch? Subscribe to ml-india's google group to join the discussion and recieve updates, news and resources about India's ml-ecosystem. Click here

ML-India is pleased to inform that the fifth ML-India meetup in its Bangalore chapter was held on 28 Aug 2016.

The discussion revolved around how NLP has evolved over past few years and different kinds of word representations. They had a speaker - Chhaya Methani, an Applied Scientist at Microsoft and former Lead Researcher at Siemens. At Microsoft, Chhaya is working on Query Reformulation and User Intent Understanding. Her areas of interests include ML, Data Analytics and Computer Vision. She gave a talk on multiple vector representations of words. She explained how most NLP approaches work with a word as the basic building block. But, since the meaning of a word can change depending on other words around it, it becomes important to take the word’s context into account. For example, The word “bank” has different meanings in the following sentences: “I deposited money in the bank”, vs “His friends could always bank on him”.

They discussed how to find similar documents by constructing term-document matrix (document ids point to word representation), word-context matrix (representing word in the context it appears in) and pair-pattern matrix (contains relationship between pairs of words). They talked about when to use word-context matrices vs pair-pattern matrices. The crowd was really enthusiastic and engaged.

The slide deck presented by Chhaya can be found here.

The discussion then moved on to the overall flow of linguistic processing and how all the processes take place in vector space models. A small discussion went around on PMI, PPMI and statistical methods used for similarity, and how NN can be used to learn word representations (word to vector) (matrix smoothing & SVD)

The response and feedback from the participants were very encouraging! They had around 30 people participating in the meetup. The participants were very active and had a lot of questions on the day’s agenda.

We’ll keep up with having such engaging meetups on ML, and having researchers and practitioners involved in ML to discuss their work and insights. Looking forward to future meetups. Spread the word to your friends from Bangalore to enroll for the meetups.

Click here to join the group.

If you wish to start a chapter of ML-India in your city, we would be more than happy to help you out and get it started. Write us a mail!