Aspiring Minds and AI

Aspiring Minds has been doing machine learning, aka artificial intelligence, for 8 years now, much before it became a vogue. We solved original questions using AI, not copycatting the West – doing a lot of firsts in the world. Here is a quick recap of Aspiring Minds’ tryst with AI, together with how AI evolved in India.

Phase I- “ML in a niche”- We had hired two engineers to work on Machine learning projects in 2010. After one year, they came to my room and inquired about their future, since all their friends were doing software development. Hardly anyone knew about ML.

- 2012: Launched SVAR: An AI based spoken English evaluation product

Today, SVAR is used across the world including India, Philippines, China and Latin America. It automatically generates scores on pronunciation and fluency based on speech samples of a person.

Was this the first AI-based product from India that reached scale?

- 2012: Made one of our data set public and organized a Machine Learning Competition

The competition had entries from India, Brazil, Belgium and Pakistan. See the leader-board and winners hereThis was probably the first by an Indian company and among the first few in the world.

- 2013: Launched AUTOMATA: World’s first machine learning based programming assessment

Automata is used by companies across the world – some examples include Wipro, Cognizant, Baidu, ZTE and one of the largest ecommerce giants in the USA. It is backed by three publications and several patents.

- 2014: Published our first ML paper on grading programming skills automatically at KDD

The paper has quickly garnered 28 citations. This was followed by several other papers on automatic grading of spoken English, motor skills, soft-skills, published at KDD, ACL, Ubicomp, IJSE and others. We also did a first workshop on AI+assessments at KDD with international collaborators.

Aspiring Minds remains one of the very few Indian companies that publish in ML conferences

 
Aspiring Minds and AI-03
 

Phase II- Big Data Science Fascination- Everyone by now had started talking about Big Data and Data Science – a new name for machine learning! Most work in India was around data engineering and not deriving intelligence from data. MOOCs on AI exploded –everyone who took the course didn’t really learn.

- 2015: Organized the world’s first Data Science Camp for Kids

We organized a very successful hands-on data science camp for kids of standard 5th-8th. Kids performed the full flow of supervised learning. Since then this open source project has been replicated at Illinois, Seattle, Pune and Bangalore. It also led to a paper on the pedagogy of teaching machine learning to kids.

- 2015: Launched ml-india.org, the first effort ever to audit India’s ML activity and a resource repository for all MLers

ML India brings all ML efforts in India under a single roof. Read more about how does India fare in ML – the main motivation behind setting up this forum. The group has 1800+ members, hosted 27 machine learning meetups, lists 146 ML professionals, 55 companies, 28 data sets and 11 groups.

We also launched a new data set AMEO, at iKDD. Attracted users from Harvard Kennedy School, Dublin Institute of Technology, New York University, TCS, Sapient and Flytxt.

- 2016: Launched the World’s first automated motor skills test

Aptitude tests have been automated for ages. But motor skills test, a way to measure skills of blue collar workers have not. We used the power of tablets and machine learning to do it and show that it is predictive of the job performance of blue collar workers. Read here.

- 2016: US Skill Map and India Skill Map- Big Data Analysis

Automatically crawled the web to aggregate jobs of USA and India, to create the world’s first interactive Skill Demand Map. Check it out here.

Phase III- National Interest in AI, but with nascent understanding –Data science had by now died a silent death, only to be replaced by Artificial Intelligence. From the PMO, Finance Minister to Niti Aayog, today everyone is interested in AI. Yet, we have little novel methods or application of AI from India. We have little local expertise in AI – our research contribution is 1/15th of US and 1/8th of China.

- 2017: Machines started understanding codes that do not compile!

Automata, our program skill grading platform started scoring uncompilable codes, a first in the world! Our algorithm could read meaning of programs, which a compiler couldn’t and generate feedback for so many more students.

And the journey continues!!!

This has been possible by efforts of many in Aspiring Minds’ research team, most notably, Shashank Srikant, Rohit Takhar, Vishal Venugopal, Gursimran Singh, Bhanu Pratap Singh, Vinay Shashidhar, and Milan Sachdeva.

Phase IV: How can India lead in Artificial Intelligence? From doing research, we started thinking about research policy. My recent book, ‘Leading Science and Technology: India Next’ focuses primarily on the research ecosystem in India and highlights several areas where we should improve. It is supported by a white paper on how India should invigorate its Artificial Intelligence ecosystem. This is where we need to go next…

- Varun Aggarwal

AI powered coding platform ‘understands’ programs that do not compile!

If you ever took a coding job test on a machine, you will probably frown if you couldn’t make your code to compile. Your program might be almost right, but due to some silly bug, unidentified in a small time frame, you will get a ZERO.

 

syntax_error

 

Not any more! Aspiring Minds’ research team has created a technology which can detect how good the program’s algorithm is, even if it doesn’t compile.

How do we do it? First, we can fix some of the codes using artificial intelligence. By looking at patterns in good compilable codes, our algorithms minimally modify existing programs to make them compilable. By using this approach, we can compile 40% of uncompilable codes. Once compilable, our patented machine learning based algorithm can generate a grade which mimics human raters.

Fancy as it may seem, we had a harder problem to solve. What about the codes that do not compile? Using smart static analysis of codes, we are able to derive features, signatures of the logic of the program, from these codes automatically. With these features and a customized form of our machine learning algorithm, we can provide grades as accurately as you could think!

On a set of programs attempted for a job in a large e-commerce player in USA, we find that 46% codes were not compiling, but weren’t blank.

Our AI based algorithm found that 6% of these codes, for 596 students, had nearly correct logic. Another 29% candidates, with a little bit of guidance, would have reached the right logic. All these candidates deserved a shot with the company!

In another data set of a technology giant in China, we find that 27% candidates whose codes do not compile, have sound programming logic.

What is more? Our AI algorithm can provide feedback to all candidates whose code do not compile. To some, we can tell how to fix their programs and make them to compile. To all, we can give them feedback on their algorithmic approach, tips to reach the correct logic and provide feedback on the stylistic and maintainability issues in their code.

Disappointed with coding platforms which gives everyone a poor score and no feedback… We have corrected this for all times to come!

- Varun Aggarwal with Rohit Takhar
Learn more about Automata – our coding platform.

Scaling up machine learning to grade computer programs for 1000s of questions in multiple languages

Machine learning has helped solved many grading challenges – spoken english, essay grading, program grading and math problem grading to cite a few examples. However, there is a big impedance in using these methods in real world settings. This is because one needs to build an ML model for every question/prompt – for instance, in essay grading, a different model designed to grade an essay on ‘Socialism’ will be very different from one which can grade essays on ‘Theatre’. These models require a large number of expert rated samples and a fresh model building exercise each time. A real-world practical assessment works on 100s of questions which then translates to requiring 100s of graders and 100s of models. The approach doesn’t yield to be scalable, takes too much time and most of the times, is impractical.

In our KDD paper accepted today, we solve this challenge quite a bit for grading computer programs. In KDD 2014, we had presented the first machine learning approach to grade computer programs, but we had to build a model per problem. We have now invented a technique where we need no expert graded samples for a new problem and we don’t need to build any new models! As soon as we have around a few tens of ‘good’ codes for a problem (automatically identified using test case coverage and static analysis), our newly invented question-agnostic models automatically take charge. How will this help us? With this technology, our machine learning based models can scale, in an automated way, to grade 1000s of questions in multiple languages in a really short span of time. Within a couple of weeks of a new question being introduced into our question pool, the machine learning evaluation kicks in.

There were couple of innovations which led to this work, a semi-supervised approach to model building:

  • We can identify a subset of the ‘good’ set automatically. In the case of programs, the ‘good set’, codes which get a high grade, can be identified automatically using test cases. We exploit this to find other programs similar to these in a feature space that we define. To get a sense of this, think of a distance measure from programs identified as part of the ‘good set’. Such a ‘nearness’ feature would then correlate with grades across questions irrespective of whether it is a binary search problem or a tree traversal problem. Such features help us build generic models across questions.

  • We design a number of such features which are invariant to the question and correlate to the expert grade. These features are inspired by the grammar we proposed in our earlier work. For instance, one feature is how different is an unseen program from the set of keywords present in the ‘good set’; while another is the difference in the programs in the kind of computations they are doing. Using such features, we learn generic models for a set of problems using supervised learning. These generic models work super well for any new problem as soon as we get our set of good codes!

Check out this illustrative and easy-to-grasp video which demonstrates our latest innovation.

 

The table presents a snapshot of the results presented in the paper. As shown in the last two columns, the ‘question-independent’ machine learning model (ML Model) constantly outperforms the test suite based baseline (Baseline). The claim of ‘question-independence’ is corroborated by similar and encouraging results (depicted in last three rows) obtained on totally unseen questions, which were not used to train the model.

Metric
Question Set
#Questions
ML Model
Baseline
Correl
All questions
19
0.80
0.65
Bias
All questions
19
0.24
0.35
MAE
All questions
19
0.57
0.85
Correl
Unseen questions only
11
0.81
0.65
Bias
Unseen questions only
11
0.27
0.31
MAE
Unseen questions only
11
0.59
0.84

What does this all mean?

  • We can really scale ML based grading of computer programs. We can continue to add new problems and the models will automatically start working within a couple of weeks.
  • These set of innovations apply to a number of other problems where we can automatically identify a good set. For instance, in circuit solving problems, the ones with the correct final answer could be considered a good set; this can similarly be applied to mathematics problems or an automata design problem; problems where computer science techniques are mature to verify functional correctness of a solution. Machine learning can automatically then help grade other unseen responses using this information.

Hoping to see more and more ML applied to grading!

Varun

Work done with Gursimran Singh and Shashank Srikant

An Automated Test of Motor Skills for Job Prediction and Feedback

We’re pleased to announce that our recent work on designing automated assessments to test motor skills (skills like finger dexterity and wrist dexterity) has been accepted for publication at the 9th International Conference on Educational Data Mining (EDM 2016).
Here are some highlights of our work –

  • The need: Motor skills are required in a large number of blue collar jobs today. However, no automated means exist to test and provide feedback on these skills. We explore the use of touch-screen surfaces and tablet-apps to measure these skills.
  • Gamified apps: We design novel app-based gamified tests to measure one’s motor skills. We’ve designed apps to specifically check finger dexterity, manual dexterity and multilimb co-ordination.
    amultifingermanual

 

 

 

 

 

 

 

 

  • Validation on three jobs: We validated the scores from the apps on three different job roles – tailoring, plumbing and carpentry. The results we present make a strong case for using such automated, touch-screen based tests in job selection and to provide automatic feedback for test-takers to improve their skills!

If you’re interested in the work and would like to learn more, please feel free to write to research@aspiringminds.com

What AM Research told you in 2015 – the data science way?

As the year came to an end, we looked back on what we shared with the world in 2015. As data nerds, we pushed all our blog articles in to an NLP engine to cluster them to identify key themes. Given the small sample size and challenges to find semantic similarity in our specialized area, we waded through millions of unsupervised samples through deep learning with a Bayesian framework, ran it on a cluster of GPUs for a month…yada yada. Well, for some problems it is just that humans can do things easily and efficiently; so that is what we actually did.

The key themes were:

Grading of programs – 4 posts

We need to grade programs better to be able to give automated feedback to learners and help companies hire more efficiently and expand the pool considered for hiring. We at AM dream to have an automated teaching assistant – we think it is possible and will be disruptive. Thus we dedicated 4 of our posts on telling you about automatically grading programs and its impact.

The tree of program difficulty – We found that we could determine the empirical difficulty of a programming problem based on the data structures it uses, the control structures used and its return type, among other parameters. We used these features in a nice decision tree to predict how many test takers would answer the question correctly, and we predicted with a correlation of 0.81! This tells us about human cognition, helps improve pedagogy and also helps generate the right questions to have a balanced test. And this is just the tip of the iceberg. Second, we approached the same by looking at the difficulty of test-cases and their inter correlation. We understood what conceptual mistakes people make and also got a recipe to make better test cases for programs and had insights on how to score them. For instance, we found that a trailing comma in a test case can make it unnecessarily difficult!

Finding super good programmers – Given these thoughts on how to construct a programming test and score it, we showed you how all this intelligence put together with our super semantic machine learning algorithm, we can spot 16% good programmers missed by test case based measures. Additionally, we also found automatically the super good ones writing efficient and maintainable code. So please say a BIG NO to test case based programming assessment tools!

venn

Reproduced from “AI can help you spot the right programmers”. It shows a test case metric misses 16% good programmers. Furthermore AI can help spot 20% super good coders

Pre-reqs to learn programming - Stepping back, we tried determining who could learn programming through a short duration course. We found that it was a function of a person’s logical ability and English but not did not depend on her/his quantitative skills. Interestingly, we found that a basic exposure to programming language could compensate for lower logical ability in predicting a successful student who could learn programming. A data way to find course prerequisites!

Building a machine learning ecosystem – 3 posts

Catching them young! We designed a cognitively manageable hands-on supervised learning exercise for 5th-9th graders. We helped kids, in three workshops spread across different cities, make fairly accurate friend predictors with great success! We think data science is going to become a horizontal skill across job roles and want to find ways to get it into schools, universities and informal education.

“Exams. I would take my exam results, from the report card of every year. And then I will make it on excel and then I will remember the grades and the one I get more grades I will take a gift” [sic.]

flashcard

Reproduced from datasciencekids.org. Whom will you befriend? Can machine learning models devised by high school kids predict this?

The ML India ecosystem – Our next victims were those in universities. We launched ml-india.org to catalyse the Indian machine learning ecosystem. Given India’s very low research output in machine learning, we have put together a resource center and a mail list to promote machine learning. We also declared ourselves as self-styled evaluators of machine learning research in India and promise to share monthly updates.

Employment outcome data release – We recently launched AMEO, our employability outcome data set at CODS. This unique data set has assessment details, education and demographic details of close to 6000 students together with their employment outcomes – first job designation and salary. This can tell us so much about the labor market to guide students and also identify gaps – to guide policy makers. We are keenly looking forward to what wonderful insights we get from the crowd! Come, contribute!

Pat our back! – 3 posts 

blog4-image

Reproduced from “Work on spoken English grading gets accepted at ACL, AM-R&D going to Beijing!”. We describe our system that mixes machine learning with crowdsourcing to do spontaneous speech evaluation

We told you about our KDD and ACL papers on automatic spoken English evaluation – the first semi-automated automated grading of free speech. We loved mixing crowdsourcing with machine learning – a cross between peer and machine grading – to do super reliable automated evaluation.

And then our ICML workshop paper talked about how to build models of ‘employability’ – interpretable, theoretically plausible yet non-linear models which could predict outcome based on grades. More than 200 organizations have benefited by using this model in recruiting talent and they do way better than linear models!

Other posts

On the posts off these three clusters, we told you about –
Why we exist – why we need data science to promote labor market meritoracy

– The state of the art and goals for assessment research for the next decade (See ASSESS 2015)

Our work on classifying with 80-80 accuracy for 1500+ classes

It has been an interesting year at AM, learning from all our peers and contributing our bit to research, while using it to build super products. We promise to treat you with a lot more interesting stuff in open-response grading, labor market standardizing and understanding next year. Stay tuned to this space!

Varun