BPLight-CNN: A Photonics-based Backpropagation Accelerator for Deep Learning Article Swipe

MAKE IMPACT

Dharanidhar Dang , Sai Vineel Reddy Chittamuru , Sudeep Pasricha , Rajarshi Mahapatra , Debashis Sahoo · YOU? · · 2021 · Open Access · · DOI: https://doi.org/10.48550/arxiv.2102.10140

Training deep learning networks involves continuous weight updates across the various layers of the deep network while using a backpropagation algorithm (BP). This results in expensive computation overheads during training. Consequently, most deep learning accelerators today employ pre-trained weights and focus only on improving the design of the inference phase. The recent trend is to build a complete deep learning accelerator by incorporating the training module. Such efforts require an ultra-fast chip architecture for executing the BP algorithm. In this article, we propose a novel photonics-based backpropagation accelerator for high performance deep learning training. We present the design for a convolutional neural network, BPLight-CNN, which incorporates the silicon photonics-based backpropagation accelerator. BPLight-CNN is a first-of-its-kind photonic and memristor-based CNN architecture for end-to-end training and prediction. We evaluate BPLight-CNN using a photonic CAD framework (IPKISS) on deep learning benchmark models including LeNet and VGG-Net. The proposed design achieves (i) at least 34x speedup, 34x improvement in computational efficiency, and 38.5x energy savings, during training; and (ii) 29x speedup, 31x improvement in computational efficiency, and 38.7x improvement in energy savings, during inference compared to the state-of-the-art designs. All these comparisons are done at a 16-bit resolution; and BPLight-CNN achieves these improvements at a cost of approximately 6% lower accuracy compared to the state-of-the-art.

Type

preprint

Language

en

Landing Page

http://arxiv.org/abs/2102.10140

PDF

https://arxiv.org/pdf/2102.10140

OA Status

green

Cited By

1

References

33

Related Works

10

OpenAlex ID

https://openalex.org/W3132993759

Raw OpenAlex JSON

OpenAlex ID

https://openalex.org/W3132993759

Canonical identifier for this work in OpenAlex

DOI

https://doi.org/10.48550/arxiv.2102.10140

Digital Object Identifier

Title

BPLight-CNN: A Photonics-based Backpropagation Accelerator for Deep Learning

Work title

Type

preprint

OpenAlex work type

Language

en

Primary language

Publication year

2021

Year of publication

Publication date

2021-02-19

Full publication date if available

Authors

Dharanidhar Dang, Sai Vineel Reddy Chittamuru, Sudeep Pasricha, Rajarshi Mahapatra, Debashis Sahoo

List of authors in order

Landing page

https://arxiv.org/abs/2102.10140

Publisher landing page

PDF URL

https://arxiv.org/pdf/2102.10140

Direct link to full text PDF

Open access

Yes

Whether a free full text is available

OA status

green

Open access status per OpenAlex

OA URL

https://arxiv.org/pdf/2102.10140

Direct OA link when available

Concepts

Deep learning, Backpropagation, Computer science, Benchmark (surveying), Convolutional neural network, Speedup, Artificial intelligence, Inference, Artificial neural network, Photonics, Machine learning, Computer engineering, Parallel computing, Geodesy, Physics, Geography, Optics

Top concepts (fields/topics) attached by OpenAlex

Cited by

1

Total citation count in OpenAlex

Citations by year (recent)

2023: 1

Per-year citation counts (last 5 years)

References (count)

33

Number of works referenced by this work

Related works (count)

10

Other works algorithmically related by OpenAlex

Full payload