[Turkmath:3461] Atılım Üniversitesi, Matematik Bölümü Genel Semineri

[Burcu Gulmez Temur]

Dear colleagues,

You are cordially invited to the general seminar organized by the
Department of Mathematics, Atılım University.

Our speaker is  *Hande Alemdar*.

 The title of her talk is " *Ternary neural networks for energy efficient
AI applications* *".*

Date: *December 12*, *2018*

Time: *15:40*

Place: *FEF 404*

Please find the abstract of the talk below.

With my best regards,

On behalf of Seminar Committee,
Burcu Gülmez Temür

*Abstract:*
Deep neural networks have achieved state-of-the-art results on a wide range
of artificial intelligence tasks. However, the computation and storage
requirements are usually quite high. This issue limits their deployability
on ubiquitous computing devices such as smart phones, wearables and
autonomous drones. In this talk, I will present ternary neural networks
(TNNs) in order to make deep learning more resource-efficient. A TNN is a
discretised neural network where the weights and activations are constraint
to three values only (-1,0, and 1). Due to this extreme limitation, there
is no standard training procedure for TNNs. I will introduce a novel
teacher-student approach for training TNNs without compromising too much
accuracy. Next, I will describe our purpose-built hardware architecture for
TNNs and present benchmark results that demonstrate up to 3x better energy
efficiency than the existing solutions.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://yunus.listweb.bilkent.edu.tr/pipermail/turkmath/attachments/20181206/f0338269/attachment.html>