Tuesday, September 11, 2018

Engineering College Department Improvements

Background
Being an industry representative of several Departmental Advisory Board (DAB) meetings, I am often presented the achievements of the department (in an Engineering college) and then asked to give suggestion on ways they can improve the overall performance of the department. Most often there are other acclaimed representatives from academic institutes in addition to the departmental staff and the college alumni, current students and in some cases the parents of the student. It is a healthy mix of in terms of age range and also the stake holders. Generally these meetings go on for about 120 minutes.

The Need
While it is possible that DAB meetings are mandated by the AICTE it does benefit the department if ofcourse the ideas get implemented or accepted by the management especially because these implementations have a cost implications.

My suggestions (not structured on purpose)

  1. Be more visible on the social media. While there is possibility of being criticized by the students especially the one with a grouse. A kid who wants to join the institute today needs more easy to understand information on the social media. Get yourself a facebook, twitter handle, may be a youtube channel, instagram, ... This way your will reach out to more of your future students.
  2. Involve your alumni. Set up a alumni network get them involved. They know the shortcoming of the department and they will speak it out with a genuine intent of making their almamater better. Sometime they might end up sponsoring internships for your students. Win-win.
  3. Always find metrics to measure your success. It is one thing to present the work done by the department and it another thing to measure it. How much can the success of the departmental events (holding a conference, workshop, cracking a hackaton, etc) be attributed to the students and how much due to the involvement of the faculty. Most of the success should be attributed to the faculty so that you can repeat the success. 
  4. Good students will definitely lead to good results, with or without the involvement of the faculty members. So always measure how many -not-so-good-students turned out to be good students at the end of four years. This measure speaks of the work put in by the faculty members of the department. Which means you have a geared up faculty members who can turn out good students. 
    1. The message is to analyze the performance of all the students and not just the top three or four.
    2. It is also important to deep dive to find out about the few students who might have failed (most colleges I go to have a very high pass percentage) - because if you figure this out you would have found the root cause and address it.
  5. It is one thing to have a library and have xyz number of books (just because AICTE mandates!) but it is best to measure its use in terms of what is being used and how much of the facuility is being used.
  6. When you measure the performance also look for temporal data. This will give you a feel for how you have been doing in terms of time. If you did well this year but it was not as good as the previous year ... may be there is something to analyze.
  7. Give self-time to the faculty in the department. While a few might misuse it .. the majority will come out with ideas that will help improve the functioning of the department. Might be able to allow them to plan research funding etc.
  8. Aim to get better ranked students. If your intake is via an entrance exam then the best measure of a departmental success is the entry and exit rank. This is a good metric of the performance of the department. If your ranks are getting fatter .. then you are doing something incorrect.
  9. Get students to collaborate and insist on multi-departmental projects. Try and involve external people (say from industry) during selection of projects. In some way the need of the industry gets embedded into the student project
  10. If publishing conference papers is a criteria (which often is!) then form a faculty group and select projects and supervise them with an aim that each project results in a publication. 
  11. Engage senior retired professors so that the young faculty in your college will have some support to fine tune their projects proposals so that they are funded (Funded projects is a big deal in most colleges)
  12. Get your students to present their projects. Video record and youtube them. These will act as a source of information for first and second year students as well as will play a good role in getting your future students interested in your department
  13. Plan student projects that can have an impact. Make sure the project continue from one batch to another
  14. Have someone be on the look out for what other colleges are doing, learn from them, it is alright to imitate them if it can result in your performance improvement 
  15. And more .... 

All these suggestions are of some use if the management is genuinely interested in growing its institute, the faculty is  committed and the students are not averse to learning in  a classroom when so much digital courses and data is available.

Endnote
The suggestions will always be incomplete ;-)

Friday, September 07, 2018

External Examiner of a Masters Thesis

Background
I often receive requests for examining thesis (mostly Masters in Engineering). Nothing to do with me, but to do with  my current job profile!

Note
This my incomplete list of Do's and Don't. Mostly written in one sitting.

Why am I writing it?
After having had to go through some painfully lengthy presentations which spoke more about the literature and less of what the student actually did!  And having had to listen to several supervisor say "I am not actually the project supervisor and I had to step in because of " to avoid any responsibility of what is being presented! I thought these tips will help the student plan in advance and speak their best.

Since I interact with the supervisor who organizes the evaluation. I will write as if I am addressing the supervisor; but the student should, they were to read, get the intent. 

Few Suggestions

Please have the student do the following 
  • Start with 10 minutes demonstration if there is something to be shown as part of the work {this give a good perspective to the thesis}
  • A strict 30 minutes presentation {at best 15-18 slides}
    • The first slide should contain the duration of the project and the number of hours spend on the project (example, Jan 2018 - July 2018; xxx hours) {this again gives a good perspective of the time spent in the project}
  • Start with the problem statement and highlight the main contribution of the work done {this should allow the student to think what they have contributed to the overall literature}
  • Concentrate on the work done rather than the literature (they could have the literature as backup slides - which they can refer to if required - but not to be presented)
  • Make sure the presenter knows what ever is written on the slide {there is no point in putting details that the presenter does not know, often happens with literature}
  • Ask them to spend a slide or two (and time) on the experimental results and analysis 
    • let them use graphs {when applicable}, 
    • highlight what they want to talk about in the tables; 
    • clearly articulate what they want the listener to gain from the tables|charts
  • Numerical if shown should be limited to two digits after the decimal {do not use 0.56892; use instead 0.57 to make it human readable}
  • They should be familiar with the mathematical expression on the slide - let them not put it without understanding 
  • They should proof read their thesis and the slides used for presentation
    • No more than 4-5 sentences in a slide
    • Do not read what is there on the slide {the examiner can do it too}
    • Use bullets instead of complete sentences
    • Time your presentation {a rehearsal  definitely helps}

Thursday, June 14, 2018

Unwanted? and Unknown? Callers == SPAM

Last Updated: June 14, 2018
Created: June 14, 2018

Just trying to get the numbers (prevalent in India) that call you endlessly and when you call back the go silent after a ring or someone want to sell you something or ...
  • Call you several times and when you call back you get to hear someone and the call drop after a second
    • 09729924389
  • Calls asking for donation | credit card | ...
    • 09136698565
    • 09930984309
    • 08291894962

Monday, March 26, 2018

Automation ↔ Jobs


Automation ↔ Jobs

 

(On a Lighter Note!)
The often debated question is will excessive automation result in greater unemployment. There have been a lots of opinion on both sides of the natural threshold, with the optimistic opinion being that “automation has always been there and what has been observed that depending on the situation people very quickly up-skilled them selves and found employment” and the pessimistic opinion being “automation is going to result in a great number of people being unemployed”. The truth probably laying somewhere in between! Automation is surely going to have an effect on people but it might not he as bad as it is being pictured! Likely Scenario (shallow view)!
Less Likely → All jobs taken over by Machines | AI | automation
More Likely → Most jobs assisted by Machine | AI | Automation
Most Likely → The jobs that humans really want to outsource to machines!
A quick look around shows that jobs can be classified into one of these few buckets
  • Teaching → coaching → tutoring
  • Exploration → Scientific → Inventions → planning → design
  • Transportation → Food, people (train | taxi | ship | airplane)
  • Cleanliness | Maintenance → Home help – cleaning
  • Food → Agriculture, distribution, preparation
  • Health → doctor, nursing, operations, health check
  • Build → living spaces, infrastructure
  • Entertainment → plays, music, movies, games
  • Security → nation, personal, cyber, policing
  • Finance → Wealth creation
  • Environment → Welfare, earth|moon|Mars, people
and it is something that we can also classify as being (a) repetitive, (b) need based, (c) new, (d) dangerous, (e) fun (f) tiring or manual and (g) that requiring empathy.
Clearly not all the jobs that humans do today are going to be taken away by machines.
Most likely, the current jobs that are repetitive, systematic, mundane are the ones that are easier for a machine to be taught | learn so that the machine can perform and hence invariably are the ones at stake (take-it-by-machine). And then there are jobs that are dangerous, tiring which humans might want to willing offload it to machine without a second thought. So the kind of jobs that are at stake, which we will see go from humans to machine are of the type repetitive, systematic, mundane, dangerous, tiring (give-it-to-machine).
Now what percentage of the current set of jobs that we do are in this “take-it-by-machine” and “give-it-to-machine” is something that is a matter of statistics which I did not care to check, but my guess is it is a large percentage depending on which part of the (developing, developed, under developed) world you are from. From the developed country perspective the human from the developing economy are the machine or automating (humanoid!) that we are talking about. So in some sense automation is already taking away jobs, it is not difficult to sense that the “economics of the job” is the aspect that is moving jobs from developed to developing shores.
Point to be noted is that there might be jobs that machines can do (meaning humans can train a machine to do the job) but they will only be taken away by machines if they can do it faster and it costs less to operate them. Automation is going to get human jobs only when they can do it correctly (error free leaning!) and economically (cost lesser than the cost incurred when a human does the job).
What is the cost of a machine to do a job? Well it includes intellectual property cost + electricity to run the algorithm + infrastructure to maintain the machine + development cost + machine maintenance cost + people cost of people who look after this job doing machine. As a consequence automation is going to go get the human job only if they can do it economically. Remember that a true businessman will usually look beyond short term. So machine will replace if and only if they are economic over a long period of time.
I believe most machine will turn out to be more expensive than human in the long run (proof?). So the only jobs that will get taken by the machines (in all probability) are the ones that the humans want to willingly give it away, the ones that are dangerous, tiring etc.

To Think About
  • What if humans could “feed directly” on solar energy – would we still care about jobs?
  • What work that I do today that I would like to outsource to a machine?
    • If there was a clause “once outsourced you can not undo it” would I still outsource it?

What can I see in Mumbai?

What can I see in the city of Mumbai?

Why read this?

Often people like to visit the city of Mumbai, either from another place within India or from outside India. Am sure there are a lot of guiding websites out there which are probably more well researched than this blog. Nevertheless when recently a friend of  mine called to check what he could show his son inside three days without traveling outside Mumbai, it set me thinking especially since I have myself stayed in Mumbai for a long time now, but frankly  havenot seen much of Mumbai myself.

My List of things to see or do?

In no particular order as yet! And no explanation on why you should visit or see them, just pointers ;-)
  1. Mahalakshmi Temple
  2. Siddivinayak Temple
  3. Flim City (you can book a tour)
  4. Helicopter ride to see the sealine of Mumbai
  5. Nehru Science Center
  6. Nehru Center Planetarium
  7. Nehru Center - Discovery of India
  8. Worli Sea Link
  9. Matheran (at a distance from Mumbai; Toy train)
  10. Karnala Bird Sanctury (at a distance from Mumbai)
  11. BEST musuem
  12. Railway Musuem (CSTM)
  13. IIT Bombay | Powai Lake
  14. Juhu beach (though it is not very clean!)
  15. Elephanta caves (boat ride) 
  16. Marine Drive (especially in the night)
  17. Gateway of India
  18. TajMahal Hotel (if you can afford, pick up a coffee!)
  19. Mumbai Darshan (BEST A/C bus; Full Day)
  20. Sanjay Gandhi National Park (Kanheri Caves)
  21. Byculla Zoo (Penguins are an attraction)
  22. Mohamad Ali Street (Pick up some food!)
  23. Linking Road (Bargain Shopping)
  24. Gandhi Musuem (Mani Bhavan)
  25. Fashion Street (Bargain clothing)
  26. Haji Ali (Dargha)
  27. Kamala Nehru Park
  28. Essel World, Water Park
  29. Hanging Gardens
  30. Chor Bazar (Flea Market)
  31. Chota Kashmir (New Mumbai)
  32. RK Studio (Chembur; not sure if there is an entry though)
  33. Local Train Ride (anywhere to anywhere -- to experience the crowd!)
  34. Metro ride (if you have not travelled in a metro)
  35. Monorail ride (between Wadala and Chembur)
  36. IMAX Wadala Dome (catch a movie)
  37. Taraporewala Acquarium

Tuesday, February 06, 2018

10 Questions to Myself! Spoken Language & Written Script

Did not do any "GoBiYa!" search!
  1. Can we define closeness of two written language scripts? Something like Telugu, Kannada; Tamil, Malayalam; Bangla, Assamese pairs seem to look similar and hence are closer to each other compared to say Telugu and Tamil.
  2. What is the origin of written script? Was it defined by the writing material (sharp pen, blunt pen) and written material (flexible, rigid) so that certain languages used similar but not exactly the same script?
  3. What is the first thing a  child draws? Is it the same when given (a) chalk and slate, (b) crayon and paper, (c) stick and sand? Do all children draw the same thing? (straight line or zig-zag lines)
  4. Why some have good looking handwriting compared to others? Is this dictated by the fineness of motor control? 
  5. Why is A (in English) and "अ" (Devanagari) the first alphabet? Is it because this the easiest and most natural sound for the speech production mechanism to make? independent of language this seems to be true.
  6. What is the origin of "A" compared to "a"? Why is this not seen in Indian language script? Atleast this (capital alphabets) seems to be exploited in named entity recognition. Like wise we have "a = अ" and "aa = आ" in Hindi which is not seen in English. Note that "A"  is equal to  "a" (English) but "a = अ" not equal to "aa = आ" (in Hindi)
  7. Why are there different number of phonemes in different languages? Does it demonstrate any fineness or intelligence of the creator? meaning if there are larger number of phonemes then the "distance" between phonemes is smaller (sound more alike) and to capture this "sound resolution" requires more fineness!
  8. Which is the latest (human) language created? Why was it created? How was it created? and When was it created?
  9. Is it possible to have a universal language? Leaving alone the politics around a language and the champions of a language do not play a role?
  10. Which is the easiest spoke language to learn? Why? Or are all languages equally easy or hard?

Tuesday, January 09, 2018

What makes a Machine Intelligent? Human Intelligence!


Background
Before we can speak if machine intelligence we need to talk a little about human intellegence. In our day to day life an othewise mundane task that is performed smartly would immediately earn the performer of the task a tag of intelligence. In some sense intelliegence is relative, contextual and even dependent on the person’s age and other attrributes. For example, if a child is able to open a closed box by unhooking the latch, the child could be attributed as being intelligent, however the same task performed by an adult would not attract any intellectual tag. So in some sense, a task performed by a person when it is not expected of him because of his qualification or his age or any other personal trait is often envisaged as intelligence. Subsequently, the same task can be performed by different humans in different ways – even if they belong to the same “category” - in such an event the most optimized way in which the task is done is often thought of as being more intelligent than the other ways of completing the same task. In effect when everything else is same it is relative else it is a function of the attribute of the hman performing the task. In some sense the definition of intelligence is very vauge and highly subjective and dependent on the relationship between the task and the human performing the task.

If we assume the methodology adopted by the human to complete a task being correlated to intelligence or on other hand a methodology that has high entropy could be considered as being more intelligently done that the methodology that is commonly used (low entropy). Clearly, the tasks that are done easily and almost involuntarily by a human are not easy for mechines to acomplish and when a machine is able to replicate what a human would otherwise do it without blinking an eye could be termed a machine intelligence.

From very early on researchers have been chasing the holy grail of building machines that can in a given scenario and given condition do exactly what a human would do. For example consider the task of opening a door with a key. More or less the machine could be programmed to perform the task of inserting the key in the keyhole and then turing it anticlockwise to open the lock. Now instead of giving one key if a bunch of keys are given to a human, most often the human tries to insert each key and keeps track of which key he used to try with all the keys one by one until he is able to open the door. A scenario change form one key to several key in a bunch could leave a machine in confusion but not a human. Going further if the task has to be performed in the dark, the human immediately uses his other cue, namely the sense of touch to find the key hole and use the sense of touch to move from one key to another to identify which key can open the lock. Clearly this changed scenario of one key -> multiple keys -> multiple keys in darkness does not require any learning by the human, it more or less comes naturaly. However a machine has to be thaught different strategies for different scenarios to achieven the same task in addition one has to make sure there are multiple sensors (camera, touch sensor) which are readily accessible to the machine.

In brief, machine is as intelligenet as a human can make it using his intelligence! So if we can programtically push in the straightforward (to the human) methods into the machine so that the mchine performas more of less the way the human performs in different scenario, the machine is considered to be intelligent. So in some sense a human has to make the machine intelligent and one can make a machine as intelligent as the human making it is and not more!

Very often the machine intelliegence is attributred to the software aspect rather than the hardware. For example, if a walking robot trips and it manages to balance itself without falling there is no acknowledgement of intelligence though there is a significant amount of programing that needs to go into making a robot to balance itself from falling. This can be attributed to the fact that this is a task, namely ability to balance when a person trips is something that comes to human very naturally. So what humans do naturally is somthing that is not attributed to a sign of intelligence when the machine is able to do it.

Making Machines Intelligent

Machines or computers by themselves are not intelligent. It is the human who needs to programatically pour in intelligence. If we suppose that machine intelligence is something that is a measure of how close the machine is able to mimic the human under similar scenarios and conditions, one is left with a question of who is the human reference? To ovecome this one uses terms like average intelligence or IQ of certain number etc – which is still very ambigious. We will not venture into this area!

So what goes into making machines “act” intelligent? Same as what makes human intellegent, learming from own (or others) experiences, learning from own (or others) mistakes. The key is learning followed by the ability to associate the learning even with a new unseen sitiation by mapping it to a similar sitiation that might have been experienced during learning. So the learning is not based on the rotting style specific to the example but is more along the direction of being able to generalize the example.

Step in machine learning algorithms which assist in allowing the machines to function intelligently. A brief taxonomy of the machine learning algorithms is shown in Figure 1.

Figure 1: Brief Taxonomy of Machine Learning.
The machine learning algorithms which make use of annotated (by human!) data (when available) are called supervised learning algorithms and the learing algorithms which do not make use of annotations (mostly because they are not available) are called unsupervised learning algorithms. The choice of the algorithm is depending on the availability of annotated data in most applications and when more than one algorithm can be used, the choice is usually based on which of these algorithms perform better.

Application

Machine intelligence is useful in all walks of life where there is a need for man-machine interaction. Consider a man machine interaction shown in Figure 2. Consider a voice based self help system run by a car insurance company and Let us consider the scenario of a customer calling the self help system to claim insurance because of an accident. Note that the Agent is chirpy when the Customer calls in and the reaction of the self help Agent could be any of the the three, namely, R1, R2 or R3.


Figure 2: Voice Based Man Machine Interaction.
Clearly, R1 is the response of the most self help system today which will in all probability result in a poor customer experience both in terms of the spoken content and the way :-) it is spoken. A speech understanding system probably would respond as R2 where in it would express empathy /Sorry to hear this./ while continuing to be chirpy :-) . While R3 which not only understands what the Customer saying but is able to empathize with the C ustomer communicate back with an emotion :-( in tune with the emotion of the Customer. Clearly, R3 kind of response will need to recognize the emotional state of the customer. So automatic detection of emotions by machine during its spontaneous conversation with a human can lead to better user experience (Ux). The ability to recognize emotion automatically makes the machine intelligent and this is something that is learnt using machine learning algorithms.

A Lesson Hidden in Two Simple Words

Sometimes profound ideas are hidden in the most ordinary places. Recently, while looking at two common English words, I stumbled upon an int...