Mi­crosoft ex­pands its port­fo­lio of ma­chine learn­ing tools

OpenSource For You - - Fossbytes -

Mi­crosoft has launched a new range of ma­chine learn­ing tools at its an­nual Ig­nite con­fer­ence. With the new launch, the Win­dows maker wants to en­able devel­op­ers to utilise easy ar­ti­fi­cial in­tel­li­gence (AI) based tools to build new ad­vance­ments.

The new tools are the Azure Ma­chine Learn­ing Work­bench, Azure Ma­chine Learn­ing Ex­per­i­men­ta­tion Ser­vice and the Azure Ma­chine Learn­ing Model Man­age­ment Ser­vice. These are de­signed to help devel­op­ers work­ing on new AI mod­els, as well as those who sim­ply want to use pre-ex­ist­ing mod­els.

The Azure Ma­chine Learn­ing Ex­per­i­men­ta­tion Ser­vice is de­signed to help devel­op­ers de­ploy ML ex­per­i­ments. Mi­crosoft has added sup­port for open source frame­works like Caffe2, PyTorch, Ten­sorFlow, CNTK and Cahiner. The ex­per­i­men­ta­tion ser­vice is de­signed to scale from lo­cal ma­chines to hun­dreds of

GPUs in the cloud.

Ma­chine Learn­ing Work­bench, on the other hand, is a desk­top client for Win­dows and Mac. The tool can act as a con­trol panel for your de­vel­op­ment life cy­cle. Lastly, the Azure Ma­chine Learn­ing Model Man­age­ment Ser­vice uses Docker con­tain­ers. Devel­op­ers can man­age and de­ploy their mod­els to any Docker con­tainer us­ing the ser­vice. The com­pany has also in­cluded its own Ku­ber­netes-based Azure Con­tainer Ser­vice within the learn­ing work­bench to en­hance the cov­er­age of its na­tive de­vel­op­ment.

All the new ma­chine learn­ing devel­op­ments are a part of Mi­crosoft’s AI plat­form, which the com­pany claims serves over 650,000 ses­sions per week.

Newspapers in English

Newspapers from India

© PressReader. All rights reserved.