Energyefficient Neural Image Processing For Internetofthings Edge Devices

• doi: 10.1109/MWSCAS.2017.8053112
• title: Energy-efficient neural image processing for Internet-of-Things edge devices
• publisher: IEEE
• isbn: 978-1-5090-6390-1
• issn: 1558-3899
• rank: 2389
• access_type: LOCKED
• content_type: Conferences
• abstract: Enhancing energy/resource efficiency of neural networks is critical to support on-chip neural image processing at Internet-of-Things edge devices. This paper presents recent technology advancements towards energy-efficient neural image processing. 3D integration of image sensor and neural network improves power-efficiency with programmability and scalability. Computation energy of feedforward and recurrent neural networks is reduced by dynamic control of approximation, and storage demand is reduced by image-based adaptive weight compression. Emerging devices such as tunnel FET and Resistive Random Access Memory are utilized to achieve higher computation efficiency than CMOS-based designs.
• article_number: 8053112
• pdf_url: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8053112
• html_url: https://ieeexplore.ieee.org/document/8053112/
• abstract_url: https://ieeexplore.ieee.org/document/8053112/
• publication_title: 2017 IEEE 60th International Midwest Symposium on Circuits and Systems (MWSCAS)
• conference_location: Boston, MA, USA
• conference_dates: 6-9 Aug. 2017
• publication_number: 8039346
• is_number: 8052834
• publication_year: 2017
• publication_date: 6-9 Aug. 2017
• start_page: 1069
• end_page: 1072
• citing_paper_count: 2
• citing_patent_count: 0
• download_count: 534
• insert_date: 20171002
• index_terms:
  • ieee_terms:
    • Energy efficiency
    • Three-dimensional displays
    • Image coding
    • Recurrent neural networks
    • Computer architecture
  • author_terms:
    • energy-efficient
    • neural network
    • image processing
  • dynamic_index_terms:
    • Image Processing
    • Internet Of Things
    • Edge Devices
    • Energy-efficient Process
    • Neural Image Processing
    • Neural Network
    • Scalable
    • Recurrent Neural Network
    • Feed-forward Network
    • Feedforward Neural Network
    • Sensor Networks
    • Image Sensor
    • Camera Sensor
    • Random Access Memory
    • Integration Of Sensors
    • Resistive Random Access Memory
    • RRAM
    • Resistive RAM
    • ReRAM
    • Storage Demand
    • Efficient Neural Network
    • 3D Integration
    • Multilayer Perceptron
    • Approximate Computation
    • Approximate Calculation
    • Synaptic Weights
    • Non-volatile Memory
    • Complex Circuits
• isbn_formats:
  • format: Print on Demand(PoD) ISBN, value: 978-1-5090-6390-1, isbnType: New-2005
  • format: USB ISBN, value: 978-1-5090-6388-8, isbnType: New-2005
  • format: Electronic ISBN, value: 978-1-5090-6389-5, isbnType: New-2005
• authors:
  • Author Name: Jong Hwan Ko
    Affiliation: School of ECE, Georgia Institute of Technology, Atlanta, GA, US
    Author URL: https://ieeexplore.ieee.org/author/37085676198
    ID: 37085676198
    Order: 1
    Author Affiliations:
    • School of ECE, Georgia Institute of Technology, Atlanta, GA, US
  • Author Name: Yun Long
    Affiliation: School of ECE, Georgia Institute of Technology, Atlanta, Georgia, USA
    Author URL: https://ieeexplore.ieee.org/author/37085888591
    ID: 37085888591
    Order: 2
    Author Affiliations:
    • School of ECE, Georgia Institute of Technology, Atlanta, Georgia, USA
  • Author Name: Mohammad Faisal Amir
    Affiliation: Georgia Institute of Technology, Atlanta, GA, US
    Author URL: https://ieeexplore.ieee.org/author/37064105900
    ID: 37064105900
    Order: 3
    Author Affiliations:
    • Georgia Institute of Technology, Atlanta, GA, US
  • Author Name: Duckhwan Kim
    Affiliation: School of ECE, Georgia Institute of Technology, Atlanta, Georgia, USA
    Author URL: https://ieeexplore.ieee.org/author/37085342980
    ID: 37085342980
    Order: 4
    Author Affiliations:
    • School of ECE, Georgia Institute of Technology, Atlanta, Georgia, USA
  • Author Name: Jaeha Kung
    Affiliation: School of ECE, Georgia Institute of Technology, Atlanta, Georgia, USA
    Author URL: https://ieeexplore.ieee.org/author/37085349274
    ID: 37085349274
    Order: 5
    Author Affiliations:
    • School of ECE, Georgia Institute of Technology, Atlanta, Georgia, USA
  • Author Name: Taesik Na
    Affiliation: School of ECE, Georgia Institute of Technology, Atlanta, Georgia, USA
    Author URL: https://ieeexplore.ieee.org/author/37085794534
    ID: 37085794534
    Order: 6
    Author Affiliations:
    • School of ECE, Georgia Institute of Technology, Atlanta, Georgia, USA
  • Author Name: Amit Ranjan Trivedi
    Affiliation: Department of ECE, University of Illinois at Chicago, Chicago, IL, US
    Author URL: https://ieeexplore.ieee.org/author/38242152200
    ID: 38242152200
    Order: 7
    Author Affiliations:
    • Department of ECE, University of Illinois at Chicago, Chicago, IL, US
  • Author Name: Saibal Mukhopadhyay
    Affiliation: School of ECE, Georgia Institute of Technology, Atlanta, Georgia, USA
    Author URL: https://ieeexplore.ieee.org/author/37278557500
    ID: 37278557500
    Order: 8
    Author Affiliations:
    • School of ECE, Georgia Institute of Technology, Atlanta, Georgia, USA

Image Sensor

• sensor_type: CMOS
• resolution: 16-Mpixel
• dynamic_range: not specified
• pixel_size: not specified
• dark_current: not specified

Optical Data

• focal_length: not specified
• aperture: not specified
• field_of_view: not specified
• distortion: not specified

Performance Metrics

• power_consumption: 3.4W

Applications & Benefits

• cell_imaging: Intelligent vision processing, used in a configured system for image classification and feature extraction.
• benefits: Programmability and scalability with improved power efficiency.

Supporting Organizations

• supported_by: Office of Naval Research, National Science Foundation Career Award.

Manuscript Details

• publication_date: 6-9 Aug. 2017

Relevancy Score

• score: 9
• missing_fields:
  • dynamic_range
  • dark_current
  • pixel_size
  • focal_length
  • aperture
  • field_of_view
  • distortion
  • frame_rate
  • signal_to_noise_ratio
  • sensitivity
  • shutter_speed
  • noise

Processed JSON Filename

request_3e7a39c9-d528-4e05-8573-c783d58252d3-energyefficient_neural_image_processing_for_internetofthings_edge_devices.json