Friday, 30 August 2019

My experience with Artificial Intelligence


In 2016, when I entered freshman year of engineering at my college, Artificial Intelligence or "A.I." as the slang went was the thing. Anyone who wasn't doing it was bound to be unsuccessful, such was the buzz. Naturally, eager to learn and excel and follow every engineer's dream of making Tony Stark a.k.a. Iron Man's A.I. system like J.A.R.V.I.S., I too picked up my laptop and books and dived right in.

While learning about the various components of A.I., the initial phase of data collection and analysis really intrigued me as it reflected with my hobby of comparing various movies, based on their all-time monetary collections, and cricket player stats. This encouraged me to take a six-week course at C-DAC, Mohali during the summer of my sophomore year in Big Data Technologies. I was so into the course at that time that I convinced the course instructor to allow me to submit the final project, which was a necessity to complete the course, without any support or team mate. So while everyone was making their projects on Big Data in teams of 4-5, I submitted a detailed report on the U.K. Traffic Accident Analysis for the year 2012-2014 which the course instructor really liked.
The detailed report is available on this blog which you can also access by clicking on the link above. Do visit the blog as it has some very interesting conclusions derived from the data, taken from the U.K. Traffic Police accident records, regarding the leading reasons of traffic accidents in the United Kingdom with great visualisations in the form of maps and graphs.

During the same time, I was also a part of a private project where we collected publicly posted data from social media sites to analyse the intent of the user. Thus, our team of four, built up a site using APIs of the social media sites and already available Natural Language Processing or NLP modules and Python's built-in libraries for data processing. Due to the Non-Disclosure Agreement, this is the extent of the information that can be provided about this project.

To understand more about the extent and developments of A.I. in the recent years, I attended a two day workshop by professional speakers on AI, ML, DL, Computer Vision and NLP. Though, the quantum of the workshop far-exceeded the time in which it was to be conducted, it surprisingly cleared many concepts and myths and opened my mind to newer techniques being used on the industry like Neural Networks and Computer Vision.

Some of this knowledge also came in handy while participating in the Smart India Hackathon, 2019 - Software Edition or SIH, when we had to make a desktop and mobile application that ranked colleges based on the students JEE/NEET/AIIMS marks and on his/her personal preference like location, hostel availability among others.

Although having had some good theoretical and hands-on experience with A.I. and M.L. in the past, my skills were put to the ultimate test while my internship at Hughes Systique, Corporation, Gurugram, where I acted as an engineering intern. My duties there involved Research and Development of a Speaker Recognition Project under the senior engineers at the corporation. I had some good time learning acoustic data collection and manipulation, Tensor Flow and the Python libraries being used to develop such tools as Microsoft Azure, Google Assistant and Amazon's Alexa.
It was there where I was introduced to CNNs or Convolutional Neural Networks. Due to the brief stay there, I was only able to get a hint of how this field of Artificial Intelligence actually works.

Currently, I am planning a support software for value investors to help ease the stresses of investing. In the software, I do plan to incorporate a system using Machine Learning that would recommend potential stocks to the investor conforming to his/her pocket and interest to maximise their gains.
I definitely do not intend to stop here and would love to work more on good projects and learn more so I could make useful products that are simple enough to be used by everybody.

1 comment: