Trip Chowdhry of Global Equities Research is known for attending industry conferences and events and providing his first-hand observations of companies under his coverage.
One type of event he attends are called "hackathons," which focus on machine learning and creation of intelligent apps and chatbots. In an e-mail to clients on Tuesday, the analyst suggested that Apple Inc. may have increased its machine learning computational budget in the past few days.
Chowdhry wrote in his e-mail that Apple increased its computational budget for Amazon.com, Inc. GPU instances by more than 200 percent to run its machine learning workloads.
Related Link:
Apple uses both G2 2xlarge (for validation and test sets) and G2.8xlarge (training sets) GPU instances. The analyst suggested that it is "not clear right now" what Apple's infrastructure for prediction is, but did state that Apple is trying to push its machine learning accuracy to above 90 percent.
Apple has many machine learning and deep learning initiatives to "make every Apple application, Apple service and Apple engagement intelligent."
"This deep learning system is powering the upcoming iOS 10 Messaging Platform, Siri Platform and part of Apple Maps. Currently, Apple has built its LSTM Algorithms on Theano Framework," Chowdhry wrote.
Did you like this article? Could it have been improved? Please email to let us know!
| Date | Firm | Action | From | To |
|---|---|---|---|---|
| Jul 2016 | Longbow Research | Initiates Coverage on | Buy | |
| Jun 2016 | Goldman Sachs | Maintains | Buy | |
| May 2016 | UBS | Maintains | Buy |
© 2016 Benzinga.com. Benzinga does not provide investment advice. All rights reserved.