In­tel In­dia Eyes AI Op­por­tu­ni­ties, Plans to Develop Ecosys­tem

Ar­ti­fi­cial In­tel­li­gence (AI) is the cut­ting edge in tech­nol­ogy. It is fast get­ting main­stream and com­ing out of the con­fines of sci­ence fic­tion. The on­set of AI-based tech­nol­ogy in In­dia is ev­i­dent in the sec­tors of e-com­merce and re­search, where en­ti­ties

Dataquest - - CONTENTS -

Igot great per­spec­tives on the po­ten­tial of AI at In­tel’s AI Day at Bangalore re­cently. I also got to know the var­i­ous parts that make up AI and why it is so com­plex. At In­tel AI Day, the com­pany deep dived into its global AI strat­egy, and high­lighted how it is work­ing with the ecosys­tem in In­dia to im­prove so­lu­tion per­for­mances on In­tel’s AI plat­forms.

Prakash Mallya, Man­ag­ing Director, In­tel South Asia, said, “As In­dia un­der­goes rapid dig­i­tal trans­for­ma­tion, the data cen­ter and the in­tel­li­gence be­hind the data col­lected will en­able the govern­ment and in­dus­try to

make ef­fec­tive de­ci­sions based on al­go­rithms. This means in­creas­ing op­por­tu­ni­ties for us­ing AI in the coun­try, and to make this hap­pen, In­tel In­dia is col­lab­o­rat­ing with Hewlett Packard En­ter­prise, Wipro, Julia Com­put­ing, and Cal­ligo Tech­nolo­gies, by en­abling them with AI so­lu­tions based In­tel ar­chi­tec­ture.”


AI is the fastest grow­ing work­load in the data cen­ter, grow­ing at two times the over­all com­put­ing mar­ket. By 2020, the in­dus­try ex­pects more servers run­ning data an­a­lyt­ics than any other work­load, and an­a­lyt­ics pre­dic­tors will be built into every ap­pli­ca­tion. Stay­ing ahead of this curve, In­tel an­nounced its AI strat­egy last year to drive break­through per­for­mance, de­moc­ra­tize ac­cess and max­i­mize so­ci­etal ben­e­fits.

To­day, In­tel pow­ers 97 per­cent of data cen­ter servers run­ning AI work­loads and of­fers the most flex­i­ble, yet per­for­mance-op­ti­mized port­fo­lio of so­lu­tions. Backed by its uni­fied ap­proach to AI, In­tel is lead­ing the AI Com­put­ing era through its end-to-end hard­ware and soft­ware that is de­signed for build­ing AI ap­pli­ca­tions across vir­tu­ally any in­dus­try.

In­tel’s col­lab­o­ra­tion with com­pa­nies such as Google, and the com­pany’s ac­qui­si­tion, in­clud­ing Saf­fron, Mo­vid­ius, Ner­vana Sys­tems and Mo­bil­eye, fur­ther In­tel’s AI ca­pa­bil­i­ties, giv­ing the com­pany an edge, es­pe­cially at a time when em­bed­ded com­puter vi­sion is be­com­ing in­creas­ingly im­por­tant the world over.

Last month, the com­pany also an­nounced a cross-In­tel or­ga­ni­za­tion – the Ar­ti­fi­cial In­tel­li­gence Prod­ucts Group (AIPG), which will align In­tel’s AI ef­forts, and rally the in­dus­try around a set of stan­dards for AI that ul­ti­mately brings down costs and makes AI more ac­ces­si­ble to more peo­ple – not only in­sti­tu­tions, gov­ern­ments and large com­pa­nies, as it is to­day. In­tel also aims to set up an ap­plied AI re­search lab ded­i­cated to push­ing the fore­fronts of com­put­ing to ex­plore ar­chi­tec­tural and al­go­rith­mic ap­proaches to in­form fu­ture gen­er­a­tions of AI. This in­cludes a range of so­lu­tions from the data cen­ter to edge de­vices, and from train­ing to in­fer­ence – all de­signed to en­able In­tel and its cus­tomers to in­no­vate faster.


AI is the com­bi­na­tion of var­i­ous fields such as ma­chine learn­ing, nat­u­ral lan­guage pro­cess­ing, com­puter vi­sion, rea­son­ing sys­tems, neu­ral net­works, deep learn­ing, depth sens­ing, pro­gram­mable sys­tems, par­al­lel com­put­ing and more.

AI is noth­ing newly in­vented, it has been around for decades. But it is only now that it is ready to take off. Kr­ish­nendu Chaud­hury, Prin­ci­pal Sci­en­tist and Head­I­mag­ing Ser­vices, Flip­kart put it well thus: “Till the mid­nineties, AI was like ‘mar­riage’— every­body’s ex­cited but prom­ises are not ful­filled. But now, it is like ‘vis­it­ing France’— it is still ex­cit­ing and it lives up to its prom­ise and there much more that can be done.”

Kr­ish­nendu points out to two rea­sons that got AI into the reck­on­ing: the big one is a mas­sive set of com­plex mod­els and tech­niques in in­for­ma­tion pro­cess­ing with a sim­ple-sound­ing name called deep learn­ing and the lesser one is the set of tech­nol­ogy de­vel­op­ments like ad­vances in hard­ware, big data and stor­age, and the open­source cul­ture that help deep learn­ing get off from con­cept to real world ap­pli­ca­tions.

Deep learn­ing is com­plex, but it solves the very im­por­tant prob­lem of ex­tract­ing ab­stract in­for­ma­tion us­ing com­plex pat­tern match­ing with neu­ral net­work anal­y­sis on huge datasets. Its ap­pli­ca­tions are myr­iad.

For ex­am­ple, Flip­kart uses deep learn­ing to give ac­cu­rate search re­sults to buy­ers who are look­ing at buy­ing a T-shirt like the one they have in mind such as a white T-shirt with the fig­ure of a ghost­like char­ac­ter or one with green and blue stripes of a par­tic­u­lar width. The de­scrip­tion of the T-shirt the shop­per wants to buy is not con­crete, it is ab­stract, hence it takes a lot of pro­cess­ing to get the re­quired re­sults. Flip­kart is sim­i­larly work­ing on deep learn­ing projects for click pre­dic­tion and sales pre­dic­tion which have di­rect busi­ness im­pact.

D N Narasimha Raju, Chief Ex­ec­u­tive Of­fi­cer, Na­tional In­sti­tute for Smart Govern­ment, who was a panelist at the AI Day, said, “Dig­i­tal trans­for­ma­tion is at the heart of govern­ment’s en­deavor to con­nect and develop In­dia. As high speed net­works get es­tab­lished, a num­ber

of dis­rup­tive changes can hap­pen. Th­ese changes will im­pact pro­duc­tiv­ity. Data an­a­lyt­ics and ma­chine learn­ing can au­to­mate pro­cesses across e-Gov­er­nance, re­search and the pri­vate sec­tor. It is rec­og­nized that the po­ten­tial of Ar­ti­fi­cial In­tel­li­gence is high in sec­tors such as health­care, dis­as­ter mit­i­ga­tion, and fi­nan­cial ser­vices. In or­der to re­al­ize this po­ten­tial, there is a need to im­part and develop the re­quired skill sets in a ma­jor way. As I un­der­stand, In­tel In­dia is en­gaged in this ef­fort and it’s proac­tive ap­proach in skill im­part­ing and so­lu­tion devel­op­ment in the area of AI by uti­liz­ing lo­cal tal­ent and col­lab­o­ra­tion will con­trib­ute sig­nif­i­cantly to­wards adop­tion of AI.”

The use cases for AI are lim­ited only by imag­i­na­tion. AI ap­pli­ca­tions range from ac­cel­er­at­ing large scale so­lu­tions, un­leash­ing sci­en­tific dis­cov­ery, aug­ment­ing hu­man ca­pa­bil­ity, and au­tomat­ing risky and te­dious tasks. The AI wave is com­ing on faster than ex­pected. AI’s im­pact will be big­ger than even the wave of dig­i­tal trans­for­ma­tion.


Ad­di­tion­ally, to en­gage stu­dents, re­searchers and de­vel­op­ers, In­tel In­dia an­nounced a com­pre­hen­sive de­vel­oper com­mu­nity ini­tia­tive – the AI De­vel­oper Ed­u­ca­tion Pro­gram, tar­geted at ed­u­cat­ing 15,000 sci­en­tists, de­vel­op­ers, an­a­lysts, and en­gi­neers on key AI tech­nolo­gies, in­clud­ing Deep Learn­ing and Ma­chine Learn­ing. Through 60 pro­grams across the year, rang­ing from work­shops, road­shows, user group and senior tech­nol­ogy leader round­tables, the ini­tia­tive aims at em­pow­er­ing this com­mu­nity with the knowhow for adop­tion of AI via tech­ni­cal ses­sions in­volv­ing the use of ready-to-de­ploy plat­forms and tools for so­lu­tion devel­op­ment.

“Our col­lab­o­ra­tion with the in­dus­try and the academia will help de­moc­ra­tize AI, by re­duc­ing en­try bar­ri­ers for de­vel­op­ers, data sci­en­tists and stu­dents. In In­dia, we are tar­get­ing the BFSI, tele­com, and e-com­merce sec­tors, across High Per­for­mance Com­put­ing (HPC), big data, and In­ter­net of Things, all of which are com­ple­men­tary to AI”, he added.

Ac­cord­ing to Pro­fes­sor Push­pak Bhat­tacharya, Director, IIT Patna, “Re­search and academia are great plat­forms to ini­ti­ate AI into the so­ci­ety. IIT Patna has been do­ing cut­ting edge re­search and devel­op­ment in Ar­ti­fi­cial In­tel­li­gence, dis­trib­uted com­put­ing, net­work se­cu­rity, so­cial net­works, and be­yond, us­ing data driven ma­chine learn­ing, as well as knowl­edge and deep learn­ing based meth­ods. Our re­search groups are cur­rently work­ing on im­ple­men­ta­tion of evo­lu­tion­ary al­go­rithms in par­al­lel en­vi­ron­ments, and us­ing In­tel based plat­forms and soft­ware tools to de­ploy, par­al­lelize and op­ti­mize sys­tems. In­tel In­dia has pro­vided IIT Patna with the nec­es­sary know-how to best op­ti­mize In­tel based AI plat­forms and tech­nolo­gies, mak­ing this col­lab­o­ra­tion ex­tremely fruit­ful and promis­ing.”

Newspapers in English

Newspapers from India

© PressReader. All rights reserved.