Search This Blog

Sunday, March 17, 2013

GATE 2013: Demand for MTech (CSE)

The result for GATE 2013 is out. The paper in Computer Science was given by 2.24 lakh students. Overall, 13.88 percent students were declared eligible for pursuing higher education. Some media reports say that for CS, a smaller fraction is eligible. One can safely assume that there would be around 25,000 students who have cleared the GATE hurdle for doing MTech in Computer Science. It is probably safe to assume that almost half of these would be more interested in jobs, IAS, MBA, going abroad, etc. So we need to take care of about 12,500 students who want to study MTech in Computer Science.

The number of students that we will be able to handle in our existing programs is just about one tenth of it. As a nation of 1+ billion people, where IT is one the largest foreign exchange earner, we do not have the resources to handle just about 12,500 students in our MTech programs.

Fortunately, this can be done overnight, at least a program of the quality similar to those MTech programs which are not in top 20.

We could start a course-based MTech program. For this purpose, a university could be set up (or an existing university could take the proposed initiative). The university would get into an agreement with top 10 institutions, say old IITs, IISc, and a coupe of more. These institutions would promise to offer just one course every semester in MOOC format. The university would let students do self-study and only conduct exams. Why would IITs agree to offer courses? Well, they are anyway thinking of doing something in the MOOC space, and this university could provide some technical/platform support and an obscene amount of honorarium. The university could also have a tie up with industry to offer such courses, and may also join hands with external platforms like coursera, etc.

Certainly not the same quality as a full-time, residential, research based program, but would be better than what is being offered by everyone other than the top 20 places.

18 comments:

2shar007 said...

Sir, I believe the ideas posted by you here are in the right spirit and is a step forward. But one thing that we in India tend to neglect is the importance of exchange of ideas in the form of discussions. My experience might be pretty limited but I have come to believe that good research requires fostering relationships that work for a greater vision. ICT in my humble opinion has yet to break ice with what human to human interaction can provide.

Diverging from the topic, wouldn't it make more sense if the biggies could act as mentors for shortlisted institutions, and also may be, organise interactions with such students in the form of Special Interest Groups. It could be a way of combining the best of both worlds, wouldn't it ?

Rahul Rai said...

Sir ,
IIT are already into the MOOC space.

The coursera class on Web Intelligence and Big Data is about to start on 25 March .


What are the IIT trying to do in the MOOC space , would you please elaborate?

The problem with Indian IT industry is like
1 . Most of the work is not original or requirng deep thinking.
2 . The work done in Indian IT space is like maintenance/suuport, which does not require deep knowledge of the things happening currently in the research space of CS.
Example: A large part of Indian IT does the work of supporting/maintaining code written in C/C++ , to interact with the OS kernel.
To get better at such things one needs to be better at reading legacy code , and knowing tricks to do bugs solving.

Is IIT willing to do such moocs?
Most of the courses in the current MOOC space are not relevant to most of the INDIAN IT jobs.
Jobs which are kind of monkey work but are hard.
I would really appreciate if IIT s do MOOC on

1. How to debug software written in Kernel Space.
2 . How to get better at finding bugs in Software and how to fix them

3 . How to test software , both written in user space and kernel space.

4 . Implementing Device Drivers for both Linux and Windows.

5. Tips and tricks in Depth for Linux add Windows.(Scripting)

You might say that these are not very instructive things to teach , but if IIT seriously want to a MOOC which would benefit the students when they work in Indian IT Industry(The outsourced one) , then there is no need to teach subjects like ML, NLP which are
1 . Already in the MMOC space
2 . Not relevant to MOST OF Indian IT .

Dheeraj Sanghi said...

@2shar007, physical interaction in terms of mentoring, etc., involves costs and egos.

Dheeraj Sanghi said...

@Rahul, Having MOOCs for improving the basic UG teaching will take care of some of the issues that you raise, but it will involve restructuring the education industry. Currently, education is provided by colleges who have no control on any aspect of academic process, and in this system, MOOCs (or NPTEL) cannot contribute much. Hence I am suggesting it as a new venture by someone who has got the regulatory permissions to offer distance education in India.

Second, courses should be taught by those who are expert in those topics and can explain things well. If the courses that IIT Professors can teach do not have a market value, they will not succeed. On the other hand, courses of the kind you are suggesting are best taught by practitioners, and it is completely alright if their courses are more popular than IIT courses.

Rahul Rai said...

It's not like that the courses on NPTEL do not have market value; the market they can influence is not much widespread in India.
The classes on NPTEL are awesome, but they are beneficial to the elite people who work on new/original projects.
But those elite already had/have the opportunity to study in the better colleges.

The goal of NPTEL , was to deliver correct/better and high quality videos to the students who study in B.Tech Shops.

But, you see the B.tech shops students end up in configuration/maintenance/support/debugging/testing/enhancement/porting type of jobs.
These jobs do not require you to master the Number/Graph theory lectures from NPTEL.

These jobs need you memorize the APIs which take in 10 arguments, these jobs do not want you know dynamic programming.

It's really essential for IIT to teach such things.

Why am I stressing on IIT?


1. It is true that such things are domain specific and are taught by practitioners in various companies or these are taught in places like NIIT/APTECH and these places charge heavy money.
If IIT would do such courses then
a) They would be free.
b) They would be more prestigious.
c) They would be serious pressure on institutes like NIIT/GNIIT/APTECH who charge tons and tons of money to teach Proxy Configuration in Red Hat! Or porting the Linux Kernel on various chips.
d) Indian IT industry gurus , who keep talking about lack of talent/knowledge in fresh college graduates would at least shut up.




2 : IIT can set up courses in their own MOOC which are taught by practitioners .
Look at how udacity got reddit co founder to teach wed-development, and people from NVDIA teach parallel programming.
With the prestige the IIT got , they can get/find practitioners/professors who would be ready to teach these things online.

It's not like that anyone can do these ventures, but IIT have to do it.
If IIT also start delivering MOOC on things like Algorithms/Probability/ML/NLP/AI just for the sake of competition with coursera , then think who would benefit from these.
The American would not benefit from these.
The Indian would not take these because the people who have high prestige in the field are already doing the same courses on coursera.

Just like IIT launched NPTEL on the LINES of MIT OCW with the motive of benefit of average Indian Engineering student, which they are not able to do currently, I am highly afraid if they launch a MOOC on say NLP, which would benefit a small part of Indian Students, and would not be the best use of effort/money/time of both the Professors/Students.


Instead if a course is launched on say OOP in depth using C ,not C++,or using C++ to write some games then these would benefit a lot of Indian Students because
A) Such courses are not there on Internet (I meant courses, not a series of tutorials).
B) This is the thing needed in most of Indian IT industry.

Hope you understand my point.





Ungrateful Alive said...

In my IIX, a typical CSE Mtech program is 166 credits, of which 90 is the Mtech "research" project. A typical MS program in CS in a top US university may be about 40 credits, of which 4 might be for independent research (equivalent to Mtech project) and the rest is regular, structured coursework. The work produced through an Mtech project is, in the best case, about 2--3 good course projects worth of work. At most 4 of 100 Mtech students produce publishable work per year in my IIX. A typical supposedly heavy graduate course (OS, networks, algorithms) in my IIX is 6 credits. In the same US university as above, it is 3 to 4 credits, but considerably harder work is needed to do well. All this means that replacing Mtech project with courses will require students to take 15 more courses, which is absurd. The current situation seems to be liked by professors and students alike. There aren't too many courses to teach. The extreme weight (90/166) on Mtech project means that no one gets bad grades in the project. Even if they did, placements are completed long before Mtech project grades are out. As Rahul Rai points out, employers are not at all interested in what students do for their Mtech projects. On the other hand, when we hire faculty members, we never test if they can teach software design, debugging, testing, or other software industry processes and practice. The first and foremost hiring criteria is visible research. So there is certainly a dichotomy between how our CSE faculty members would like to lead their lives and what value students want from my department. It is easy, from this point on, to recognize that the students, being the customers, are always right, and moreover, we are publicly funded, and therefore, we have no right to our ivory towers. To some extent, with fast-bloating class sizes and the inevitably less personal learning process, this die has already been cast. The results are already not pretty. But then, IIXs are a momentary aberration in India's history and future trajectory. It is obvious that populism will kill IIXs and what little quality remains in them. So we might as well retrain to be a Microsoft certification shop, like, yesterday.

Prashant Sahdev said...

sir gate2014 pattern is going to be changed or will remains same?

Varun Aggarwala said...

While your ideas are laudable, I still think that they are futile. There is no way that the Government can handle 12000+ students without compromising the quality of its existing programs. I am not against opening new IIT's but I think that the industry should also rise to the occasion. The reason why silicon valley is the mother of all technical innovation is not just because of Stanford or Berkeley, but because the local industry is invested in the development of new programs and facilities at these institutes. Leave aside CS, what about other streams? I feel that the pharma/biotech industry in India has a lot of potential, but they are not ready to commit resources to train the next generation of bright minds. Of course the government will open institutes and will train such folks, but at some point the industry also has to step up and get into the game.
Does the Indian IT industry want to partner with IIT's or other Engineering Colleges, to train the next generation of engineers?

Ungrateful Alive said...

Varun --- don't get tunnel-vision-ed into looking at Stanford and Berkeley. The Bay Area IT industry gets a very large number of recruits from excellent teaching colleges in the area like Harvey-Mudd and Mills College that Indians don't even know about, but their bachelor programs are far better than any IIT. Fundamentally there is no fix for the situation that a CS/IT department ranked approx 300 in USA still gives a better overall UG education than the best of the IITs.

Varun Aggarwala said...

@Ungrateful Alive: Did you smoke something, this fine morning before making this ludicrous statement?
"CS/IT department ranked approx 300 in USA still gives a better overall UG education than the best of the IITs."
I did my undergrad from one of the new IIT's and trust me, when I say it was comparable if not better than the education given at one of the Ivy League universities where I am currently studying. Is it better in all regard? No. Can we do a better job? Yes.
However dismissing all IIX's as an aberration can only imply that you are trolling.
I am very much familiar with small colleges in America like Harvey-Mudd or Swathmore. They are pretty decent, but I would choose IIT any day over them. I think the system needs reform, but it is not in shambles as you think.
Coming back to the original point about the industry partnering with the universities, I can only tell you about relevant examples from my area of Genomics/Computational Genetics. Greater Philadelphia area has lots of Pharma/Biotech companies, doing some of the coolest work possible. However they face critical shortage of scientists/staff, who can test relevant biological hypotheses using solid statistical/computational methods. They recently pledged a sum of over 10 Million USD to open a medical informatics center at Penn. I feel such initiatives can be taken by companies every where in the world. There is a nascent biotech sector in Bangalore, and I see no reason as to why they can't partner with IISC or other institutes to develop departments, who can then churn out students/research which in turn can contribute to the economy of the region.

Dheeraj Sanghi said...

@Varun, you don't have to be personal. That an IIT grad would find academic content in even well respected universities as simple does not imply that his/her "education" was much better. There can be other metrics of quality of education, such as, how much improvement an IIT brings about compared to this other university in two similarly endowed students. There can also be other measures, which include, the percent of students going for higher education, or percent of students setting up their own companies.

You may claim that all these measures are useless, but then it is your view versus someone else's view. No point in being personal for this.

Varun Aggarwala said...

I should have generously used smileys in my previous comment :) Watching way too much of South Park has transformed me into a small Cartman.
Coming back to the point, I think the focus of education in US as compared to India is different. I do not think that IIT's are better than American universities and neither do I think the other way round is true. I only find it objectionable when people make blanket statements that top 300 US universities are better than IITs. Prof. Sanghi, you did you undergrad from IITK and then your PhD from UMD. Do you seriously think that UMD provides better education than IITK? I am sure it might do a better job in some aspects, but in others IITs also do a very fine job.
I am sorry for hijacking this thread :)

Dheeraj Sanghi said...

@Varun, I used to have these discussions with my friends called, "what would you advise a Martian?" So one day we discussed the advantages of IIT education vis-a-vis UMCP education (partly perceived since I did not go through UG education at UMCP). We felt that there are three major advantages over UMCP: 1. Closer to home, in a cultural environment which we are used to. 2. A very low cost. and 3. Membership of Alumni Association (in the sense that the perceived value of membership of IITK AA was higher than value of membership of UMCP AA, at least amongst my peers).

So, what will I suggest a Martian (so culturally and physical distance, both are equi-distant), who is probably going back to Mars after his education (so, membership of AA is not important), and both IITK and UMCP are curious enough about him to offer him free tuition, lodging and boarding (so cost is not an issue).

The answer was: UMCP.

Varun Aggarwala said...

You are most definitely very funny :)
Thought experiment laced with humor; I am sure the students in your lab/class never get bored.
If cost was not an issue, and Penn was ready to give me a USD 250,000 education for free, then by all means I would choose that. However we do not live in such an ideal world.
I think of everything as an optimization problem. I feel given the constraints of cost, future avenues and mental satisfaction, I would choose IIX over any other place. However this is discounting the fact that getting into an IIX is not simple.
Also at the end of the day you are only as good as you want to be. Getting an education at UMCP or a top university might give you a head start, but ultimately life is a marathon, and I think my education at IITG has trained me well to be a long distance runner :)
I humbly apologize again for steering the discussion into a completely tangential direction. I would most certainly blame my education at IIX for this :)

Ungrateful Alive said...

I have been (literally) sniffing IIT the last 14 years, and the air is dense and heavy with it.

ears said...

Sir there is an initiative in our state 3 colleges being set up in Andhra Pradesh,by the name of "Rajiv Gandhi University of Knowledge Technologies" the M.Tech program there is offered by using MOOCS from nptel etc .....

Naren said...

Dear Sir
Your recommendation is heard by Georgia Tech and they are offering MOOC PG in CS in USD 7000: http://www.insidehighered.com/news/2013/05/14/georgia-tech-and-udacity-roll-out-massive-new-low-cost-degree-program

Naren said...

Dear Sir
Your recommendation is heard by Georgia Tech and they are offering MOOC PG in CS in USD 7000: http://www.insidehighered.com/news/2013/05/14/georgia-tech-and-udacity-roll-out-massive-new-low-cost-degree-program