Gaurav Goswami

Gaurav Goswami is a PhD student in Image Analysis and Biometrics Lab at IIIT-Delhi. Gaurav has been working mainly in the field of machine learning and deep vision and it's applications in the field of computer vision. His thesis focuses on using machine learning and computer vision techniques to solve challenges in face recognition. He has been awarded the prestigious IBM PhD fellowship for the year 2015-2016.

   Video Face Recognition

He talked about his recent project on Video Face Recognition where he has proposed a different technique of face recognition in videos from the earlier state of the art techniques. The proposed algorithm, termed as MDLFace, achieves state-of-the-art performance even at low false accept rates.

Nipun Batra

Nipun Batra is a final year PhD student at IIIT Delhi advised by Prof. Amarjeet Singh (IIIT Delhi) and Prof. Kamin Whitehouse (Univ. of Virginia). His research interests include computational sustainability and data analytics. He received his B.E. in Computer Science from Delhi College of Engineering in 2011, where he worked on UAVs. His PhD work has been well received by the community in the form of various awards- best PhD forum presentation at SenSys 2015 and best demonstration at Buildsys 2014.

   Towards practical energy disaggregation

Buildings, specifically residential homes, across the world contribute significantly to the overall energy consumption. A lot has been studied on how to reduce the energy wastage in the homes. Several sensor deployments have been used in the past to understand and improvise residential energy consumption. Of particular interest is appliance energy feedback, shown to save up to 15% household energy. Deploying large number of sensors for each appliance across millions of homes is challenging from an installation and maintenance perspective and is also cost prohibitive.
More..

Dr. Tapan Gandhi

Dr. Tapan Gandi is an Assistant Professor at IIT-Delhi and a Research Affiliate at MIT. He completed his PhD in the field of Biomedical Engineering from IIT-D on Behavioral & Neurophysiological Correlates of Object Representation and Evidence of Neural Plasticity in 2011. He did his Post Doc in the Dept. of Brain and Cognitive Sciences, MIT. He has been a recipient of multiple fellowships including MIT fellowship for PhD and was selected amongst the best 30 candidates in the world to get advanced training on Neuro-Imaging at UCLA.

   Project Prakash

He talked about his work in studying kids and adults with treatable conditions of blindness and how Project Prakash answers neuroscientific questions regarding object learning and brain plasticity. He explained how the analysis of the brain's response to visual stimuli, and the development of visual skills, after eye-sight is recovered, helps in understanding the prospects and process of visual recovery after a lifetime of blindness. Read more about the work here.

Sahil Bhatia

Sahil is a computer science student at Netaji Subhash Institute of Technology. He has strong inclination toward machine learning applications and research.

   Automated Correction for Syntax Errors in Programming Assignments using Recurrent Neural Networks

Sahil presented a technique for automatically generating repair feedback for syntax errors for introductory programming problems. It provides a feedback on syntax errors that uses Recurrent neural networks (RNNs) to model syntactically valid token sequences.

Harsh Agarwal

Harsh is an independent consultant and recently completed a research study on ragging mandated by the Supreme Court. He is the co-founder of CURE, India's first and largest non-profit organisation dedicated solely towards the elimination of ragging and promotion of more positive ways of interaction between the freshers and seniors in Indian universities and colleges.

   Understanding Ragging from a Psychosocial Perspective

He answers the primitive questions related to ragging. He explains how ragging is probed using a psychosocial framework and how a psychosocial framework recognises the multiple factors, psychological and social, that need to be taken into account to understand ragging behaviour, and puts them in systemic relation to one another.

Pushpendre Rastogi

Pushpendre is a CS Ph.D. student in The Center For Language and Speech Processing at the Johns Hopkins University. Currently, he is working on solving problems posted on StackOverflow. He has published on a variety of topics such as Multiview LSA, Efficient Adaptive SPSA and Weighted FST with Neural Networks.
More..

   Neural Encoding with Structured Decoding

He explains how deep neural network can help extract the functions, from a small amount of data, to convert a string from one form to another. The DNN can be trained with a small dataset and would work on different languages. He also explains that how without doing major feature engineering as is done in the state of the art technique, his model is performing equally good.

Una-May O'Reilly

Una-May O'Reilly is a Principal Research Scientist at MIT Computer Science and Artificial Intelligence Laboratory. She leads the AnyScale Learning For All (ALFA) group. She has expertise in scalable machine learning, evolutionary algorithms, and frameworks for large-scale, automated knowledge mining, prediction and analytics.
More..

   Lab Projects

Dr. Una-May discussed three of her's lab's projects - MOOCDB, FCUBE and FLEXGP

Amarjeet Singh

Amarjeet completed his Ph.D from UCLA where he worked at the intersection of robotics, statistics, machine learning, ecology and biology. He was an Assistant professor at IIIT Delhi until recently and is now the CTO of zenatix
More..

   Energy Analytics

Electrical consumption, for both residential and industrial consumers, is currently provided through monthly bills that give little insights into our consumption patterns. This data if collected at a higher resolution - say every minute, can provide deep insights. In this talk I will discuss how we use data to help our customers reduce on their energy spend by 10%.

Tarun Cherukuri

Tarun Cherukuri is the co-founder and CEO of Indus Action. An alum of BITS Pilani, he gave up his high paying job at Hindustan Lever to dedicate himself to the education of underprivileged children. After spending 2 years as a Teach for India fellow in Pune, Tarun was awarded the coveted Fulbright Scholarship to pursue his Masters at Harvard Kennedy School, Massachusetts, USA.
More..

  Project Eklavya

He talked about Project Eklavya and gave an overview on Right To Education Act Section 12(1) (c), importance of social diversity and the idea of Inclusive India. He also mentioned about his aim to nurture a million inclusive schools in India by 2050

Sameep Mehta

Sameep Mehta is a Senior Researcher and Manager at IBM Research - India. He received his Ph.D. in Data Mining and Visualization from Ohio State University. His current research interests are Data Mining, Text Mining, Machine Learning, Big Data Technologies, Social Data Analytics and Knowledge Graph.
More..

  Knowledge Graph and Data Platforms

He talked about his work at IBM Research - India and gave an overview of graph analytics and various data platforms like Titan, Apache Spark etc.

Nisha Thomson

Nisha Thompson is the co-founder of DataMeet, an NGO founded in Bangalore. DataMeet gets technologists, open data enthusiasts, government officials, bureaucrats and journalists under one roof and gets them to talk about data, open data and civic issues.
More..

  DataMeet's journey and overview of India's open data situation

She talked about DataMeet's journey from ideation to where it currently stands and gave us an overview of India's Open Data situation and the nuances involved in such policy/attitude changes.

Kalyan Veeramachaneni

He is a research scientist at MIT's famed Computer Science and Artificial Intelligence Laboratory (CSAIL), where he works at the intersection of big data, machine learning, and data science, and co-heads a group called Anyscale Learning For All. More..

  Mining the big data from Massive Open Online Courses MOOC's

Kalyan will give an overview of big data, discuss his role and activities in the field, and explain how he hopes to significantly improve our ability to predict adverse health events, estimated wind power, and student behavior on online platforms. He will give a number of examples of systems built to make predictions and then deep dive into a very specific domain that is geared towards MIT’s mission of - educating a billion people. More..

Dibyendu and Kireet

Dibyendu and Kireet are final year masters students at IIITD. Their work was presented at CoDS on how to intelligently use social media to solve today's traffic problems. Their work was also selected among the winners of data challenge at CoDS.

  TrafficKarma: Estimating Effective Traffic Indicators using Public Data

Dibyendu and Kireet joined us for a talk on 17th April 2015 regarding their work presented at CoDS. More...

Aditya Johri

Aditya Johri, Ph.D. is an Associate Professor in the Applied Information Technology Department. Dr. Johri studies the use of information and communication technologies (ICT) for learning and knowledge sharing, with a focus on cognition in informal environments. More...

  Education technology and its state of the art.

Aditya will lead a discussion on education analytics and the state of the art in it. More...

Rishabh Singh

Rishabh Singh is a researcher at Microsoft Research, Redmond. He completed his PhD at the Computer Science and Artificial Intelligence Laboratory at MIT. More..

  Program Synthesis for the Masses

New computing platforms have greatly increased the demand for programmers, but learning to program remains a big challenge. Program synthesis has the potential to revolutionize programming by making it more accessible. More..

Vijay Ganesh

Vijay Ganesh is an assistant professor at the ECE Department of the University of Waterloo, Canada. More..

  SAT and SMT Solvers for Software Engineering and Security

Boolean SAT and SMT solvers increasingly play a central role in the construction of reliable and secure software, regardless of whether such reliability/security is ensured through formal methods, program analysis or testing. More..

Manas Mittal

Manas Mittal is a PhD student at the computer science department at UC Berkeley. More..

  How much we should pay to crowdsourcing workers to maximize the amount of work done for a given budget?

Every day millions of crowdsourcing tasks are performed in exchange for payments. Despite the important role pricing plays in crowdsourcing campaigns and the complexity of the market, most platforms do not provide requesters appropriate tools for effective pricing and allocation of tasks. More..

Saugato Datta

Saugato Datta is a Vice President at ideas42. He works with partners to design, test and scale programs and products that use behavioral economics to benefit poor people in developing countries. More..

  Behavioral Design - Applying Behavioral Economics to Social Problems

The Knowledge Session will look at a number of ways in which behavioral economics helps us redefine problems, arrive at fresh diagnoses for why they happen, and then design and test innovative solutions.

Anirudh Krishna

Dr. Krishna is a professor of public policy and political science at Duke University. More..

  The Higher-Variance Society: The Other Half of India in an Age of Globalization

An insightful discussion on various factors affecting the Indian Economy and how it should be governed. The Indian economy, hailed for nearly two decades as a growth miracle, has recently slowed down, falling into a dip. More..

Short duration visitors

Samarth Singal is an undergraduate at Harvard University, majoring in computer science and economics. He interned at Aspiring Minds' research lab in the summer of 2015. When he's not busy with his coursework, he enjoys reading and hiking in the Cambridge area.

Arun Saigal is a software engineer at Quizlet, where he works on both Android and web development. A recent graduate of MIT, Arun brings a longtime commitment to edtech, including three years on App Inventor (an Android application to teach app development to students).

Maya Escueta is a Policy and Training Manager for CLEAR/J-PAL South Asia at IFMR. She works on building capacity for monitoring and evaluation and bridging the gap between the findings of J-PAL’s research and actionable policy in the region.

Samarth Mohan is an undergraduate at MIT majoring in Electrical Engineering and Computer Science. He interned at Aspiring Minds Research & Development Center for a period of two months in 2016.​ He helped in developing a MOOC on applied machine learning. He is interested in Cricket and Music. Before Aspiring Minds, he was an intern at Akamai.

Artificial intelligence and its application for educational assessment and recruitment

Varun Aggarwal, ZTE Corporation, Shenzhen, China, 30 November, 2016
Hosted by: Dong Hexing and Wei Shunbao (ZTE).

Artificial intelligence specifically machine learning is disrupting things around us in image recognition, product marketing and even driving cars. We will begin by discussing some of these examples and then specifically talk about AI's application to build automated assessments, link assessment results to job success and understand the labor market. I will discuss examples of automatically evaluating computer programs, free flow spoken English, motor skills and soft skills. We use machine learning, crowdsourcing and exploit the touch interface to develop these tests and scoring mechanisms. We will see how results of such assessments predict job performance aided by interpretable models. Finally, I discuss how large scale automated assessment and employment outcome data could provide labor market insights. All these technologies together help scale learning and drive meritocracy in the labor market.

References:
a."A system to grade computer programming skills using machine learning.", S. Srikant and V. Aggarwal, KDD 2014
b. "AutomaticSpontaneous Speech Grading: A Novel Feature Derivation Technique using the Crowd.", V. Shashidhar, N. Pandey, V. Aggarwal, ACL 2015
c. "Learning Models for Personalized Educational Feedback and Job Selection", V. Shashidhar, S. Srikant, V. Aggarwal, Machine Learning for Education, Workshop at ICML 2015
d. "Question Independent Grading using Machine Learning: The Case of Computer Program Grading", G. Singh, S. Srikant and V. Aggarwal, KDD 2016
e. "Knowing what not to do is a criticzal Job Skill", S. Stemler, S. Nithyanand, V. Aggarwal, IJSE, 2016
f. "Apps to measure motor skills of vocational workers", B.P. Singh and V. Aggarwal, Ubicomp, 2016.


Automated Grading, Certification and Labor Market Insights using data science

Varun Aggarwal, Signals, Inference and Networks (SINE) Seminar, University of Illinois Urbana-Champaign, 17 November, 2016

Learners need real-time feedback on their assignments/test attempts to enhance learning. They also need credentials to demonstrate their learning to employers to get matching jobs. Automated and semi-automated assessments are key to scale learning and delivering its socio-economic benefits.

We use AI techniques to build automated assessments, link assessment results to job success and understand the labor market. In this talk, I will discuss examples of automatically evaluating computer programs, free flow spoken English, motor skills and soft skills. We use machine learning, crowdsourcing and exploit the touch interface to develop these tests and scoring mechanisms. We will see how results of such assessments predict job performance aided by interpretable models. Finally, I discuss how large scale automated assessment and employment outcome data could provide labor market insights. All these technologies together help scale learning and drive meritocracy in the labor market.

References:
a."A system to grade computer programming skills using machine learning.", S. Srikant and V. Aggarwal, KDD 2014
b. "AutomaticSpontaneous Speech Grading: A Novel Feature Derivation Technique using the Crowd.", V. Shashidhar, N. Pandey, V. Aggarwal, ACL 2015
c. "Learning Models for Personalized Educational Feedback and Job Selection", V. Shashidhar, S. Srikant, V. Aggarwal, Machine Learning for Education, Workshop at ICML 2015
d. "Question Independent Grading using Machine Learning: The Case of Computer Program Grading", G. Singh, S. Srikant and V. Aggarwal, KDD 2016
e. "Knowing what not to do is a criticzal Job Skill", S. Stemler, S. Nithyanand, V. Aggarwal, IJSE, 2016
f. "Apps to measure motor skills of vocational workers", B.P. Singh and V. Aggarwal, Ubicomp, 2016.


Industry Expectations and Skill Development

Varun Aggarwal, Engineers Conclave, IIT Madras, 2016, 1-3 September, 2016

What skills are required by the industry? Do students in engineering pick up skills that the industry requires? What are the different kinds of skills required for different job roles? Do skilled students end up getting job matching their skills?

We will answer these questions using a data based approach: analysis of scores of AMCAT, an employability assessment, of a million of engineering students and their employment outcome. We will discuss the proportion of candidates who are employable for different jobs, what skills are deficient and what biases prevent job-ready candidates from getting a job. For instance, data shows that only 8% engineers are employable for core engineering jobs, while only around 3% have skills for a software engineering job in an IT product company. Through these insights, we will discuss what kind of skill building efforts are required and also touch upon some of the common problems with skill development programs. We will look in to the skills required for new kind of jobs and also, how automation and AI is changing the job world. Finally, we will discuss how an assessment led recruitment ecosystem can help provide information and make the entry-level recruitment ecosystem much more efficient.

About the event: The Engineers' Conclave is an annual mega event conducted by the Indian National Academy of Engineering in partnership with different reputed institutions providing a platform for engineers and engineer-scientists to address some of the major challenges in engineering under two chosen tracks. More details here.


Uncovering ability and personality to maximize people success and efficiency – An AI approach

Varun Aggarwal, Indian Institute of Technology, Delhi, 24 August, 2016
Hosted by: Prof. Tapan Kumar Gandhi

People have different abilities, personality and soft skills which help them learn different things and do different jobs efficiently. Without doubting the infinitude of the human potential, if we could understand who can do what jobs/trainings well and match them, people will come become more satisfied and there will be larger economic efficiency. We use principles of psychology and AI to uncover this for millions of candidates every year.

In this talk, I will discuss examples of using machine learning and crowdsourcing to automatically evaluate programming, motor skills and spoken English skills. I will discuss couple of examples of what skills make someone do well in performing certain jobs. We will build interpretable machine learning models and also, analyse what behaviors make someone successful in the workplace. Finally, I will introduce couple of our ecosystem building efforts: www.ml-india.org and www.datasciencekids.org


Data science startup eco-system in India

Gursimran Singh- Panel member at KDD 2016, San Francisco, August 2016
Other panelists: Anand Rajaraman (Co-founder, Cambrian Ventures), Abinash Tripathi (Co-founder, Helpshift), Parul Gupta (Co-founder, Springboard).
Moderator: Manish Gupta (Xerox Research)

Today India is becoming the new leader of Startup ecosystem across the world. These starts are enabling people to solve their problems through highly sophisticated technical solutions. With the latest success in data science, nowadays many Indian start-ups are building in-house research and data science teams to leverage the power of data and tackle the problems with the different angle. A panel of experts from emerging startups was called at KDD to discuss these problems and how data science is being used in practice​ to solve these problems.

Interesting points were discussed. Almost everyone agreed that data science is paramount for India to tackle the problems at scale and deliver services at affordable cost. For instance, a gentleman in the audience emphasized the need of creating machine learning based auto-tutoring systems which can enable quality and affordable access to education in India. The panel also discussed examples of optimizing logistics using data science. The panel and audience agreed to the need to strengthen data science ecosystem in India by floating indigenous conferences and initiatives. See more details here.


AI for Automated Educational Assessments and Labor Market Insight

Varun Aggarwal, Hong Kong University of Science and Technology, 25 April, 2016
Hosted by: Prof. Pascale Fung (http://www.ece.ust.hk/ece.php/profile/facultydetail/pascale)

Learners need real-time feedback on their assignments/test attempts to enhance learning. They also need credentials to demonstrate their learning to employers to get matching jobs. Automated and semi-automated assessments are key to scale learning and delivering its socio-economic benefits.

We use AI techniques to build automated assessments, link assessment results to job success and understand the labor markets. In this talk, I will discuss examples of using machine learning and crowdsourcing to automatically evaluate computer programs and free flow spoken English. Then I discuss an example of connecting results of such assessments to job performance by building interpretable models. Finally, I discuss how large scale automated assessment and employment outcome data could provide labor market insights. All these technologies together help scale learning and drive meritocracy in the labor market.

References:

  1. A system to grade computer programming skills using machine learning.", S. Srikant and V. Aggarwal, KDD 2014
  2. Automatic Spontaneous Speech Grading: A Novel Feature Derivation Technique using the Crowd.", V. Shashidhar, N. Pandey, V. Aggarwal, ACL 2015
  3. Learning Models for Personalized Educational Feedback and Job Selection", V. Shashidhar, S. Srikant, V. Aggarwal, Machine Learning for Education, Workshop at ICML 2015
  4. http://research.aspiringminds.com/whitepaper-automated-assessments/



ML in Program Analysis

Shashank Srikant and Varun Aggarwal, Guest lecture, IIT Madras, 8-9 Feb, 2016:
Hosted by: Prof. Rupesh Nasre (www.cse.iitm.ac.in/~rupesh)

A three hour guest lecture in Prof. Rupesh Nasre's 2016 graduate-level course on Program Analysis (www.cse.iitm.ac.in/~rupesh/teaching/pa/jan16/schedule/). Taught a class of 30 on how machine learning can be applied in program analysis to predict insights on program behavior.


AI for automated educational assessments and labor market insight

Varun Aggarwal, Supercomputer Education and Research Center, IISc, Bangalore, 5 Feb 2016
Hosted by: Prof. Partha Pratim Talukdar (http://talukdar.net/))

Learners need real-time feedback on their assignments/test attempts to enhance learning. They also need credentials to demonstrate their learning to employers to get matching jobs. Automated and semi-automated assessments are key to scale learning and delivering its socio-economic benefits.

We use AI techniques to build automated assessments, link assessment results to job success and understand the labor markets. In this talk, I will discuss examples of using machine learning and crowdsourcing to automatically evaluate computer programs and free flow spoken English. Then I discuss an example of connecting results of such assessments to job performance by building interpretable models. Finally, I discuss how large scale automated assessment and employment outcome data could provide labor market insights. All these technologies together help scale learning and drive meritocracy in the labor market.

References:

  1. A system to grade computer programming skills using machine learning.", S. Srikant and V. Aggarwal, KDD 2014
  2. Automatic Spontaneous Speech Grading: A Novel Feature Derivation Technique using the Crowd.", V. Shashidhar, N. Pandey, V. Aggarwal, ACL 2015
  3. Learning Models for Personalized Educational Feedback and Job Selection", V. Shashidhar, S. Srikant, V. Aggarwal, Machine Learning for Education, Workshop at ICML 2015
  4. http://research.aspiringminds.com/whitepaper-automated-assessments/


Data Science in Indian Startups

Panel member at iKDD CoDS 2015 - Varun Aggarwal, Bangalore, March 2015

See details here.


Applying Machine Learning to Industry problems

Varun Aggarwal, Nasscom ATC, December 2014

With the surge in data that's generated by the second all around us, using machine learning is a natural step forward to make sense of it. Besides conventional domains such as search and recommendation, machine learning is a potent tool to learn more on more subjective domains such as training, matching and hiring. In this talk, we will discuss two such examples: prediction of employment outcomes and what makes someone successful in a short duration programming training.


Democratizing Job Opportunity

Varun Aggarwal, MIT, USA August 2014

Do people, on completion of higher education pick up skills that the industry requires? If a person is employable, what is the likelihood of him/her getting a job? MOOCs create access to education, but do they create access and equity in getting jobs? The answers to these questions are critical for an efficient labor market and a healthy meritocratic society. More..


KDD: ASSESS 2014

Industrial Panel at ASSESS 2014, KDD
Moderated by Varun Aggarwal, Bloomberg LP, USA, August 2014
Panelists: Piotr Mitros (edX), Lav Varshney (UIUC), Collin Fuller (Khan Academy)

Learnings from a Million Employability Assessments

Varun Aggarwal, Keynote, QUT Senior Leadership Group Conference, Australia, May 2014; Microsoft Research, USA, April 2014
Vinay Shashidhar, Shashank Srikant, Microsoft Research, India, June 2014; Xerox Research Center India, June 2014

An assessment-driven job marketplace could drive accountability in higher education and meritocracy in labor markets. Towards this goal, we have conducted over a million assessments of job candidates, matched candidates to jobs based on them, and tracked their success. By mining this large scale data, we learn how to make the employment ecosystem more efficient while discovering heretofore unknown trends. More..


Grading computer programs using machine learning

Varun Aggarwal, Queensland University, May 2014; Microsoft Research, April 2014

The automatic evaluation of computer programs is a nascent area of research with a potential for large scale impact. Extant program assessment systems score mostly based on the number of test-cases passed, providing no insight into the competency of the programmer. More..


Aspiring Minds: An Overview, Industrial Track Presentation

Shashank Srikant, ACM DEV 2013, Bangalore, India, January 2013

See talk overview here. SimProg is now Automata.


Learning for HR Decision Making and Social Impact

Varun Aggarwal, IBM TJ Watson Research Center, March 2011; CSAIL, MIT, USA, March 2011

Human Resource management is generally considered a soft and subjective science. We look at HR decision making, in particular, matching a candidate to a job profile, from the standpoint of statistical learning. We discuss a theoretical framework to lay out the problem and identify the statistical learning problems involved. Given the theoretical assumptions and business requirements, the models developed need to affirm to certain mathematical properties such as coordinate-wise monotonicity, sparsity and interpretability. More..


Equity and fairness in employment market in India

Varun Aggarwal, MIT India Reading Group, MIT, USA, March 2011

The talk will discuss the employment market for entry level job seekers in India. Primary topic of discussion shall be inefficiencies in the job market and how it is impacting both job seekers and employers. The two key strands of the discussion shall be: equal opportunity to job for entry level job seekers and identifying candidates with high propensity of success for a given job. More..


Statistical Learning and Prediction @ Aspiring Minds

Varun Aggarwal, Industrial Session, SEAL, India, December 2010 (Varun also chaired the Industrial session)


Word2Vec

Manya Wadhwa [PPT]

We touched upon the basic idea behind word2vec. Talked about the two models included in this concept, along with a small demo. Concluded with a discussion on the applications of this concept to real world text analysis problems.



Apache Spark

Saqib Nizam Shamsi [PPT]

We explored Apache Spark and how it is used in the field of Big Data. We discussed its design principles, the abstraction called Resilient Distributed Datasets (RDD) which Spark uses, and compared it to Hadoop in terms of performance.



Hadoop

Sameer Saini and Tarun Sahni [PPT]

We talked about what is Big Data and how Hadoop provides a solution to Big data problems . We also had a overview of Hadoop File System(HDFS) and Map Reduce. We also used a 3 Node Hadoop Cluster to execute a Map Reduce job.



Mongo Database and NoSQL Databases

Adarsh Kumar [PPT]

We discussed about Nosql databases and MongoDB, its usecase, various concepts like sharding and replica sets and its comparison with sql on various parameters.



Static Code Analysis

Saqib Nizam Shamsi [PPT]

We discussed about the motivation behind and the need for static analysis specifically in the determination of code quality. In the light of the same, we briefly explored concepts in programming languages like regular expressions, syntax trees, data flow graphs and how we use this body of knowledge in the Stylistic module of Automata. [Paper]



Different Types of Network Attacks

Tarun Sahni [PPT]

We discussed about various online attacks like Cryptanalytic Attacks, Injection Attacks, Phishing, DoS, Spoofing and Malwares. We talked about the different sub-types and their consequences and ways to detect some of these attacks.



Asymmetric Key Encryption Techniques (Diffie Hellman)

Gursimran Singh [PPT]

This discussion revolved around cryptographic encryption and decryption techniques including symmetric and asymmetric key exchanges and an extended talk on Diffie Hellman algorithm. [Paper]



Neural Networks

Ashutosh Pandey [PPT]

In this talk we discussed about neural network training using back-propagation, regularization in neural nets and its extension to deep neural networks.



Recommendation systems

Harsh Nisar [PPT]

Discussed broadly about collaborative filtering, content based recommender and SVD based approaches. Followed up with a literature survey and a discussion about the differences/similarity between these algorithms and traditional prediction problems.



Game theory

Harsh Nisar

Touched the basic principles of Game theory, discussing concepts like game, pay-offs, dominant and dominated strategy, Prisoners Dilemma, Nash equilibrium and coordination games. Concluded with everyone playing the classic Guess the 2/3 of the average game.



Learnings from iKDD, CODS 2015

Gursimran Singh

Learnings and brain pickings from CODS, iKDD conference held at Bangalore.



Python data ecosystem

Harsh Nisar

The talk covered the state of the art in the python data ecosystem with respect to visualizations and data wrangling and its usefulness at work. Libraries like sklearn, pandas and searborn were covered. The power of IPython notebooks in meetings and EDA was also covered.