Machine Learning Competition
Know more about Aspiring Minds' Research here.
The ML CUP results has been announced. Congratulations to the winners. Click here to view the results.
Philosophy: Computer science and engineering to solve real problems
At Aspiring Minds, we are interested in using computer science and engineering to find solutions to real-world problems. We are specifically intrigued by problems which address an important need of the human race, our society and which identify a solution that shall make the world a better place to live in. Honestly, problems such as how to get more users to spend time on a social networking website or making a viral game do not tickle our imagination. Purpose-driven and always driven by a good purpose, we have just one more criterion to choose our problems of interest: They should be hard. Easy problems are boring!
The Goal: Solving the problem of meritocracy in jobs
One such problem is efficient matching of people to jobs by merit: one that appeals most to the worker while delivering maximum value to their employer. We believe that there is a job for every individual. We wish to identify that job, match the individual to that job and facilitate the employment of the individual. On the other hand, we want to assist companies in acquiring talent which most closely matches the job requirement. We have a set of twin goals: equal job opportunity and meritocracy.
We not only strongly feel but have data to show that current ways of job-matching are deficient. They leave many employable candidates without jobs and companies without the right set of people to do the job. The solution to this problem will empower the ten million strong youth of India to steer their career by merit and contribute towards the growth of our thousand billion dollar economy.
We think that this is a computer science problem, more specifically, a pattern recognition problem. Strange? Well such is the world of science.
The Science: Looking for patterns in human abilities
We are big fans of pattern recognition or what some call machine learning. We look for patterns in data which help us 'predict'. To be precise, we write programs to look for patterns in data which help us predict outcomes. If we look close enough, we can find rather odd and interesting patterns in novel places: is there a pattern among people who form the best of computer programmers vs. the rest? Can we predict whether a particular person will enjoy a particular job? Or can we predict that a particular person shall be a great truck driver, follow rules and not cause accidents...
We are not satisfied with being able to predict an outcome, we are keen to understand what these patterns 'mean'. Are they just some set of complicated numbers and mathematical functions or can we understand why a particular number or a function predicts a particular outcome.
Let us take an example. Suppose we are to take a voice sample of a person in order to find out what was said. Our program takes as input the discretized speech sample at a given time interval, does x number of transformations, applies y mathematical functions and z pattern recognition methods and predicts accurately what was said. This solves the prediction problem but, we want to understand what does x, y and z imply. Is it the frequency of speech that is important to predict the outcome, the amplitude or something entirely different? Why is a certain function over a certain variable the right predictor and the other function is not? What was it that really made the predictor work?
Needless to say, it helps us solve our problems more effectively. More importantly, it helps us to go beyond engineering, to do science, find the cause, why of things. We believe this is a step towards achieving the larger traditional goals of artificial intelligence: to understand 'why' certain models are able to predict certain things.
Machine Learning Competition
We want to share some of the problems we regularly solve with all of you. We think the youth should be a part towards solving a problem which largely affects them.
Not only do we think you will enjoy solving a real-world problem using computer science, we also think you will come up with some innovative solutions which we shall learn new things from. And in the process, some of you will make some good pocket money! Ready to take the challenge?
Experts at Aspiring Minds and Dr. Una-May O'Reilly, EvoDesignOpt, CSAIL, MIT shall judge the final entries.