AI/Machine Learning, Bharat and Bhartiya IT Industry

The Technology & Economic Forum is a venue to discuss issues pertaining to Technological and Economic developments in India. We request members to kindly stay within the mandate of this forum and keep their exchanges of views, on a civilised level, however vehemently any disagreement may be felt. All feedback regarding forum usage may be sent to the moderators using the Feedback Form or by clicking the Report Post Icon in any objectionable post for proper action. Please note that the views expressed by the Members and Moderators on these discussion boards are that of the individuals only and do not reflect the official policy or view of the Bharat-Rakshak.com Website. Copyright Violation is strictly prohibited and may result in revocation of your posting rights - please read the FAQ for full details. Users must also abide by the Forum Guidelines at all times.
ArjunPandit
BRF Oldie
Posts: 3759
Joined: 29 Mar 2017 06:37

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby ArjunPandit » 01 Oct 2019 04:00

Supratik wrote:India's first artificial intelligence shopping store. This technology is becoming very popular in developed countries. In the next 20-30 years most routine human work are going to be done by robots. It means lesser and lesser people will be required to do such routine jobs. One more reason why India should think of stricter population control measures.

https://youtu.be/mz1GN1nnOrQ

where india can excel is commoditizing this luxury ..making these things available en masse in all areas...baki puncture shops will always be required.

ArjunPandit
BRF Oldie
Posts: 3759
Joined: 29 Mar 2017 06:37

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby ArjunPandit » 01 Oct 2019 04:15

thanks sir,
my viewpoint on new languages are
1. unless some big tech firm adopts it and releases significant library it will be hard for any language to replace python in this space
2. python has significant 'early mover' advantage through libraries/packages related to data accessibility, ML modeling, reporting and other automation. For Most FS firms getting additional system is a challenge from IT management (installation, maintanance and what not) and resource training perspective. THat reverse feeds even Unis too
3. the level of efficiency we are looking would only be required for real time large scale pricing applications. Very few would want to be the first mover in this space.

NRao
BRF Oldie
Posts: 17142
Joined: 01 Jan 1970 05:30
Location: Illini Nation

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby NRao » 01 Oct 2019 07:12

It is capable of using established libraries.

Gen programs can combine generative models written
in Julia, neural networks written in TensorFlow, and custom
inference algorithms based on an extensible library of Monte
Carlo and numerical optimization
techniques.


Bolded are all libraries.

Having said that, it is not a language in the traditional sense:

............... Gen, a new
probabilistic programming system with novel language constructs for modeling and for end-user customization and
optimization of inference
. Gen makes it practical to write
probabilistic programs that solve problems ..................


So, I prefer "modeling probabilistic situations where one can customize and optimize the outcome".

It seems to be good in the defense sector (IF one can get their hands on classified data, of course.)

ArjunPandit
BRF Oldie
Posts: 3759
Joined: 29 Mar 2017 06:37

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby ArjunPandit » 01 Oct 2019 14:36

nraoji
I.
i doubt one would get defence data..however, a lot of data would be 'dual use', i provide categories and leave the rest to your imagination with a cup of tea or alcoholic beverage
1. images:
2. text: .
yes there would be lot of nuances of mapping algos trained on civilian data to military applications, as the speed, and hardware requirements would be very different, pose different constraints and opportunities.

II. Regarding the point on libraries, my bad should have clarified, i meant libraries internal to the firm. I am sure you're aware that firms customize external libraries with their own codes/macros/functors. The difference is minor but the advantages are huge for customization. i dont think python requires too much coding either..unless one is writing a modeling approach from scratch...most of the times its the usage of these libraries...

ArjunPandit
BRF Oldie
Posts: 3759
Joined: 29 Mar 2017 06:37

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby ArjunPandit » 01 Oct 2019 14:38

it might be a good exercise to try image processing to identify vikram using the LRO images around the area of interest. are you familiar with image processing?

ArjunPandit
BRF Oldie
Posts: 3759
Joined: 29 Mar 2017 06:37

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby ArjunPandit » 01 Oct 2019 14:40

^^^^
Amber G. wrote: ......

sir would it be possible for you to guide in this area..

Vayutuvan
BRF Oldie
Posts: 10881
Joined: 20 Jun 2011 04:36

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby Vayutuvan » 01 Oct 2019 23:53

ArjunPandit ji, I am not @Amber G. Would my 'guidance' count?!

ArjunPandit
BRF Oldie
Posts: 3759
Joined: 29 Mar 2017 06:37

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby ArjunPandit » 02 Oct 2019 13:34

Vayutuvan wrote:ArjunPandit ji, I am not @Amber G. Would my 'guidance' count?!

Of course sir....

ArjunPandit
BRF Oldie
Posts: 3759
Joined: 29 Mar 2017 06:37

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby ArjunPandit » 02 Oct 2019 16:21

possible ..would still like to try..

Amber G.
BRF Oldie
Posts: 7471
Joined: 17 Dec 2002 12:31
Location: Ohio, USA

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby Amber G. » 24 Jan 2020 01:48

Cyber-Physical Systems Hubs (including AI, ML, IoT, Industry 4.0, Robotics...)... are about to take shape.
State of AI in India

Vayutuvan
BRF Oldie
Posts: 10881
Joined: 20 Jun 2011 04:36

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby Vayutuvan » 25 Jan 2020 02:45

Amber G. wrote:Cyber-Physical Systems Hubs (including AI, ML, IoT, Industry 4.0, Robotics...)... are about to take shape.
State of AI in India


Can somebody define Cyber-Physical Systems? @CramS ji?

NRao
BRF Oldie
Posts: 17142
Joined: 01 Jan 1970 05:30
Location: Illini Nation

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby NRao » 25 Jan 2020 03:53

Vayutuvan wrote:
Amber G. wrote:Cyber-Physical Systems Hubs (including AI, ML, IoT, Industry 4.0, Robotics...)... are about to take shape.
State of AI in India


Can somebody define Cyber-Physical Systems? @CramS ji?


Wiki.

Language
Download PDF
Watch
Edit
A cyber-physical (also styled cyberphysical) system (CPS) is a system in which a mechanism is controlled or monitored by computer-based algorithms. In cyber-physical systems, physical and software components are deeply intertwined, able to operate on different spatial and temporal scales, exhibit multiple and distinct behavioral modalities, and interact with each other in ways that change with context.[1] Examples of CPS include smart grid, autonomous automobile systems, medical monitoring, industrial control systems, robotics systems, and automatic pilot avionics.[2]

........

jaysimha
BRFite
Posts: 1412
Joined: 20 Dec 2017 14:30

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby jaysimha » 13 Oct 2020 13:57

ISRO International Webinar on System Engineering Aspects of Artificial Intelligence in Robotics
Event Schedule – October 14, 2020 (9.30 hrs. to 19.30 hrs. IST)



https://edubard.in/full/isro-international-webinar-on-system-engineering-aspects-of-artificial-intelligence-in-robotics
Image

https://edubard.in/media/media/uploads/2020/10/12/programme-schedule_webinar141020.pdf

jaysimha
BRFite
Posts: 1412
Joined: 20 Dec 2017 14:30

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby jaysimha » 04 Dec 2020 16:32

Image
Dec 02, 2020
From its borders to cities to schools, India is made safer using AI
By Samiksha Mehra

https://indiaai.gov.in/article/from-its-borders-to-cities-to-schools-india-is-made-safer-using-ai

Kudos to UP police...

Amber G.
BRF Oldie
Posts: 7471
Joined: 17 Dec 2002 12:31
Location: Ohio, USA

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby Amber G. » 08 Dec 2020 00:30

Some paths to Quantum Computing being pursued in India with support of DST leading up to a part of upcoming Quantum Technology mission...
Warming up for speed computing
Image

ArjunPandit
BRF Oldie
Posts: 3759
Joined: 29 Mar 2017 06:37

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby ArjunPandit » 08 Dec 2020 05:51

https://www.cbsnews.com/news/mohsen-fak ... chine-gun/

A satellite-controlled machine gun with "artificial intelligence" was used in last week's assassination of a top nuclear scientist in Iran, the deputy commander of the Revolutionary Guards told local media Sunday. Mohsen Fakhrizadeh was driving on a highway outside Iran's capital Tehran with a security detail of 11 guards on November 27, when the machine gun "zoomed in" on his face and fired 13 rounds, said rear-admiral Ali Fadavi.


The machine gun was mounted on a Nissan pickup and "focused only on martyr Fakhrizadeh's face in a way that his wife, despite being only 25 centimetres away, was not shot," Mehr news agency quoted him as saying.

It was being "controlled online" via a satellite and used an "advanced camera and artificial intelligence" to make the target, he claimed.

Fadavi said that Fakhrizadeh's head of security took four bullets "as he threw himself" on the scientist and that there were "no terrorists at the scene."

Iranian authorities have blamed arch foe Israel and the exiled opposition group the People's Mujahedeen of Iran, known as MEK, for the assassination.

State-run Press TV had previously said "made in Israel" weapons were found at the scene.

Various accounts of the scientist's death have emerged since the attack, with the defense ministry initially saying he was caught in a firefight with his bodyguards, while semi-official Fars news agency claimed "a remote controlled automatic machine gun" killed him, without citing any sources.

According to Iran's defense minister, Amir Hatami, Fakhrizadeh was one of his deputies and headed the ministry's Defense and Research and Innovation Organization, focusing on the field of "nuclear defense."

Amber G.
BRF Oldie
Posts: 7471
Joined: 17 Dec 2002 12:31
Location: Ohio, USA

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby Amber G. » 08 Dec 2020 07:32

^^^ Also covered in Iran thread..

NRao
BRF Oldie
Posts: 17142
Joined: 01 Jan 1970 05:30
Location: Illini Nation

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby NRao » 08 Dec 2020 08:37

ArjunPandit wrote:https://www.cbsnews.com/news/mohsen-fakhrizadeh-iran-scientist-killed-satellite-controlled-machine-gun/

A satellite-controlled machine gun with "artificial intelligence" was used in last week's assassination of a top nuclear scientist in Iran, the deputy commander of the Revolutionary Guards told local media Sunday. Mohsen Fakhrizadeh was driving on a highway outside Iran's capital Tehran with a security detail of 11 guards on November 27, when the machine gun "zoomed in" on his face and fired 13 rounds, said rear-admiral Ali Fadavi.

.................


Two comments:

1) IF it was AI based, then there is no need for "satellite controlled". AI is sufficient to make decisions on its own
2) IF "AI" was involved, it is is truly beyond amazing

ArjunPandit
BRF Oldie
Posts: 3759
Joined: 29 Mar 2017 06:37

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby ArjunPandit » 08 Dec 2020 18:29

NRao wrote:
ArjunPandit wrote:https://www.cbsnews.com/news/mohsen-fakhrizadeh-iran-scientist-killed-satellite-controlled-machine-gun/

A satellite-controlled machine gun with "artificial intelligence" was used in last week's assassination of a top nuclear scientist in Iran, the deputy commander of the Revolutionary Guards told local media Sunday. Mohsen Fakhrizadeh was driving on a highway outside Iran's capital Tehran with a security detail of 11 guards on November 27, when the machine gun "zoomed in" on his face and fired 13 rounds, said rear-admiral Ali Fadavi.

.................


Two comments:

1) IF it was AI based, then there is no need for "satellite controlled". AI is sufficient to make decisions on its own
2) IF "AI" was involved, it is is truly beyond amazing

not necessarily..AI autonomy is not binary, there are degrees of autonomy in AI. Read the link below
https://www.rolandberger.com/publicatio ... chines.pdf

While the quoted example is for manufacturing, it is all throughout. Even though the capability for fully autonomous AI exists in almost all fields to have autonomy, the adoption has not been extensive and thus the verification results do not exist. The current line of thought is first have basics sorted out for good time. Then it will be put in place. The nuts and bolts are already there.

Returning to your point on satcom; it depends where does the matching algo run, and where does info with which it matches the real time image lie. In exact facial recog algo, the info has to lie in a server located in US and not in the machine. face matching info can't be stored in the end point of machine gun. In either case it is fascinating.

ArjunPandit
BRF Oldie
Posts: 3759
Joined: 29 Mar 2017 06:37

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby ArjunPandit » 08 Dec 2020 18:36

Amber G. wrote:^^^ Also covered in Iran thread..

sir i saw your comments there. May be the ground evidence points against the news. But I dont think that the capability to do such operations remotely is beyond the reach of capabilities. In fact all the nuts and bolts for it exist and are open to most even in public domain. Only if one is smart enough to triangulate (or more) information from various ways. I know I am talking like UB, but there is more to Tesla and palanatir stocks are not rising without any reason. Even if the evidence against it exists, the balloon hasnt been let go without any reason

Amber G.
BRF Oldie
Posts: 7471
Joined: 17 Dec 2002 12:31
Location: Ohio, USA

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby Amber G. » 08 Dec 2020 22:44

^^^ We will know soon enough as the event is fairly big and Iran will not be able to hide much if there were security lapses. Current thinking (among people I respect) is involvement of satellites (time delay among other factors) it extremely unlikely (or even cost-effective as Israel will not like to take a chance if something goes wrong and they have to be near 100% accurate etc). (FWIW My one son worked for some time in Google's autonomous car design - he later worked for another startup which also specializes in AI for such things. Lot of technology here comes from Israel (and some of the companies located there are extremely good ).. anyway we will know soon.

I did happen to see one analysis, derived from high res photo at the crash site (which shows bullet holes in cars) and using 3D modeling to locate the source or origin ..of course using non-classified data - more for fun than anything else which was interesting... will be interesting to see how good this model did in predicting when more details come out.

ArjunPandit
BRF Oldie
Posts: 3759
Joined: 29 Mar 2017 06:37

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby ArjunPandit » 10 Dec 2020 01:13

not sure why do you say that sat com would be time/cost effective that too in a car which has or can have good power supply . With regards to time lag , wouldnt the time lag in few seconds..in the end predator drones operate from us to afghanistan also, note i mentioned tesla, not google, google was more on LIDAR for autonomous cars rather than tesla, which focussed on Computer vision they seem to be far ahead of everyone ...also what is being said is that machine gun was ai controlled. Of course this is all my hypothetical thinking. I think it will be a combination of multiple sensors..if done..else our wild imagionation which could be close to real world..

Vayutuvan
BRF Oldie
Posts: 10881
Joined: 20 Jun 2011 04:36

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby Vayutuvan » 10 Dec 2020 07:00

Amber G. wrote:I did happen to see one analysis, derived from high res photo at the crash site (which shows bullet holes in cars) and using 3D modeling to locate the source or origin...

Photogrammetry?

NRao
BRF Oldie
Posts: 17142
Joined: 01 Jan 1970 05:30
Location: Illini Nation

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby NRao » 11 Dec 2020 10:36

ArjunPandit wrote:not necessarily..AI autonomy is not binary, there are degrees of autonomy in AI. Read the link below
https://www.rolandberger.com/publicatio ... chines.pdf


IF you know that military applications have "levels" (Levels on page 5 of that doc) I would be very interested in the definition. As far as I know there are none in military *combat* applications - for extremely good reasons.

ArjunPandit wrote:While the quoted example is for manufacturing, it is all throughout.


Not so.

Defense applications have ethics that is *absolute* (no compromise) primary AI, followed closely by GP and planning. All of which are not present in the commercial side as *forcefully*.

I doubt AI was used here. Too complex a situation.

This was a SF application, which is the most complex of all AI applications (from an AI point of view)

On sat: it was used 1) to give a thumbs up or down, and 2) to verify the result. I have no idea how it could have been used to control anything - too much latency. No network can keep up with AI. And, why deploy an AI solution if it cannot reconcile for itself?

ArjunPandit
BRF Oldie
Posts: 3759
Joined: 29 Mar 2017 06:37

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby ArjunPandit » 11 Dec 2020 11:18

NRao wrote:
ArjunPandit wrote:not necessarily..AI autonomy is not binary, there are degrees of autonomy in AI. Read the link below
https://www.rolandberger.com/publicatio ... chines.pdf


IF you know that military applications have "levels" (Levels on page 5 of that doc) I would be very interested in the definition. As far as I know there are none in military *combat* applications - for extremely good reasons.

ArjunPandit wrote:While the quoted example is for manufacturing, it is all throughout.

given that we are talking in general statements, let me quote one, Absence of evidence is not evidence of absence

Not so.

Defense applications have ethics that is *absolute* (no compromise) primary AI, followed closely by GP and planning. All of which are not present in the commercial side as *forcefully*.

Does not seem how this statement is relevant here. Care to elaborate?


I doubt AI was used here. Too complex a situation.
WHile i agree with complexity of situation. It is completely vague statement. Let's first establish what do you mean by AI?

This was a SF application, which is the most complex of all AI applications (from an AI point of view)
on the contrary i think SF could be the most suitable, high risk to the assets, very very strong planning element, execution does require a final permission or man in the loop system

On sat: it was used 1) to give a thumbs up or down, and 2) to verify the result. I have no idea how it could have been used to control anything - too much latency. No network can keep up with AI. And, why deploy an AI solution if it cannot reconcile for itself?


lets first agree on what all are you including in AI, and what do you mean by AI. then it will be easier to discuss/debate further. Latency is a problem..but at the same time it can can taken care of ..again i am not saying it was done in this case, all i am saying is that it is not beyond the reach and it can be done..unless you have conclusive evidence to point otherwise..

jaysimha
BRFite
Posts: 1412
Joined: 20 Dec 2017 14:30

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby jaysimha » 06 Apr 2021 14:07

FICCI e-Symposium on AI for Air Warriors
Apr 05, 2021, Virtual Platform

http://www.ficci.in/past-event-page.asp?evid=25463



neerajb
BRFite
Posts: 836
Joined: 24 Jun 2008 14:18
Location: Delhi, India.

Re: AI/Machine Learning, Bharat and Bhartiya IT Industry

Postby neerajb » 06 May 2021 12:46

Just read this thread and replying very very late. ArjunPanditji if latency aka real-time is not an issue then Python is really versatile, easy to use and has large number of libraries for virtually everything and that's why it is popular even among non-programming folks (which was the idea to start with). BUT if you want decisions made in real-time in milliseconds and repeat that process multiple times in a second then Python is not going to cut it especially with compute heavy deep learning algorithms.

I have experienced that C/C++ gives the best performance (x), followed by Java (25x-50x) and then Python (200x) for our ML workloads. Recently implemented a real-time solution for fraud detection using a compute heavy deep learning algorithm. Python was immediately out of scope as the inference alone was taking more time than the latency of the entire pipeline! C++ is kind of hard to integrate with other components so the trend is to move towards Java. Also with newer iterations Java is getting better.

Regarding performance, Most of the python libraries have python wrapper running C/C++ code under the hood. Nvidia also provides out of the box software to run on their GPUs (Triton Inference Server) which again follows the same path. Interacted with Manjeera Digital systems for our use case and they have an interesting solution for compute acceleration.


Return to “Technology & Economic Forum”

Who is online

Users browsing this forum: No registered users and 6 guests