Special Interest Group Machine Learning, IIT Kanpur (SIGML)

With the current surge in attempts to emulate human "intelligence", the importance of Machine Learning cannot be downplayed in any way. Machine Learning is the science of creating algorithms that are adaptive to data features. In simpler terms, it involves algorithms that change itself according to the data it is presented with. Take, for example, the "familiar" task of recommending products to users on an e-commerce site. A very popular approach is to study user data and find other users similar to you and then recommend the products they have looked at, to you. It is essentially identifying similar users but it adapts itself to extract the "important" features in the user data. To achieve this, the algorithm must be able to identify what is important, extract hidden features (eg. pattern in the sequence of page visits) etc. This is where Machine Learning comes in.

Machine Learning, today, is immensely capable of diving "deep" into huge volumes of data and extracting useful information. One might argue that AlphaGo defeating Lee Sedol in the game of GO is more about learning fast from huge volumes of data than about actually being more intelligent. However, nevertheless, it did defeat Sedol in the game. This capability of adapting to the data and extracting useful information is very important in environments where it wouldn't be possible for a human programmer to consider all possibilities. While AlphaGo learned to play GO, it played against itself millions of times and during each run, it learned what a good move is and what isn't. The number of possible configurations a GO game can run through is hugely many and no programmer could humanely sit down and code all these possibilities down. A good GO player relies on his intuition and ability to look at the consequences of all possible moves for a few steps ahead. It was this intuition that AlphaGo had to extract from all the million runs that it played against itself.

Similar ability to extract useful information from data and adapt itself to operate on it is seen in AI agents like Apple's Siri, Google now, Microsoft's Cortana, Amazon's Alexa and much more. Other applications today include search engines, like Bing and Google, learning to rank web pages against your query, companies, like Amazon and Uber, dynamically pricing articles and services based on historical data, demand, and other product characteristics. The list is really long. Machine Learning is a very stimulating field for people from diverse fields including statistics, algorithms, optimization, Computer hardware, economics, neuroscience, information theory and much more. The wide variety of techniques range from Neural networks (Deep Learning) to online and reinforcement learning, Bayesian learning, Kernel Learning etc. The origins of machine learning are debated but many agree that the field arose out of a need for algorithms for data processing and artificial intelligence. In its early days (1980s) when the big conferences in machine learning such as ICML, NIPS and AISTATS were just getting started, machine learning benefited from influence from diverse fields such as algorithm design, statistics, and even neurosciences. The influence of these areas can still be seen in machine learning. Notable, departments in IIT Kanpur not only offers broad general courses in Machine Learning but also specific courses into most of these topics.

SIGML, or the Special Interest Group for Machine Learning, IIT Kanpur is a group of people, who share an excitement in Machine Learning, Computer Vision, Natural Language Processing and Data Mining to discuss latest developments and research options. IIT Kanpur is evolving into a stronghold of Machine Learning expertise, with more than a dozen professors in the CSE and EE departments who specialize in diverse areas of machine learning, artificial intelligence and its application to diverse fields like Computer Vision, NLP, Robotics, Recommendation etc. It is an endeavor to bring people who share an excitement in Machine Learning, Computer Vision, NLP, and Data Mining to discuss latest developments and research options. The group aims at organizing problem-solving sessions, seminars, research days, workshops and guest lectures. SIGML was rejuvenated during the year 2015-2016 y effort from faculties & interested student. SIGML tries to encourage students to actively participate in exploring the potent world of Machine Learning and provides opportunities for the same by a wide range of activities. Group is open to both to all people inside & outside the community. The group aim at encouraging machine learning research within the institute, provide research resources and guidance to students interested in ML, provide research assistant to startup in ML and to promote industrial-academia research collaboration. Special Reading groups are also organized in various fields like Concentration Inequality, Deep Learning. Probabilistic Machine Learning & Cognitive Science etc. Apart from this invited talks by distinguished professors, research students and company experts are organized quite frequently. The group is very active and recently organize series of external skype seminar (skype / on campus) lectures on recent advances in machine learning. Many researchers were invited for various talks (including Skype talks). For more information please visit SIGML website (www.cse.iitk.ac.in/users/sigml) and Wikipedia Page (http://wiki.iitk.ac.in/mediawiki/index.php/SIGML). You can register for the mailing list by filling the form on site. If you like this, you can also check out new Vision group of IIT Kanpur (https://www.cse.iitk.ac.in/users/vision/).

Thanks to Dr. Purushottam Kar for testimony regarding future of Machine Learning

Edited by Vivek Gupta & Amartya Sanyal