fbpx

“High-fidelity Conversion of Floating-point Networks for Low-precision Inference Using Distillation with Limited Data,” a Presentation from Imagination Technologies

James Imber, Senior Research Engineer at Imagination Technologies, presents the “High-fidelity Conversion of Floating-point Networks for Low-precision Inference Using Distillation with Limited Data” tutorial at the May 2021 Embedded Vision Summit.

When converting floating-point networks to low-precision equivalents for high-performance inference, the primary objective is to maximally compress the network whilst maintaining fidelity to the original, floating-point network. This is made particularly challenging when only a reduced or unlabelled dataset is available. Data may be limited for reasons of a commercial or legal nature: for example, companies may be unwilling to share valuable data and labels that represent a substantial investment of resources; or the collector of the original dataset may not be permitted to share it for data privacy reasons.

Imber presents a method based on distillation that allows high-fidelity, low-precision networks to be produced for a wide range of different network types, using the original trained network in place of a labeled dataset. The proposed approach is directly applicable across multiple domains (e.g. classification, segmentation and style transfer) and can be adapted to numerous network compression techniques.

See here for a PDF of the slides.

Here you’ll find a wealth of practical technical insights and expert advice to help you bring AI and visual intelligence into your products without flying blind.

Contact

Address

1646 N. California Blvd.,
Suite 360
Walnut Creek, CA 94596 USA

Phone
Phone: +1 (925) 954-1411
Scroll to Top