Neuromorphic Computing in CMOS: Digital, Analog or Mixed-Signal ? - - PowerPoint PPT Presentation

▶
neuromorphic computing in cmos digital analog or mixed
SMART_READER_LITE
LIVE PREVIEW

Neuromorphic Computing in CMOS: Digital, Analog or Mixed-Signal ? - - PowerPoint PPT Presentation

Dec 28, 2022 379 likes •546 views

Neuromorphic Computing in CMOS: Digital, Analog or Mixed-Signal ? Shreyas Sen, Ayan Biswas, Priyadarshini Panda, Kaushik Roy ECE, Purdue University Sep 27, 2016 SPARC Lab 1 Neuromorphic Fundamental Question Error-Resiliency

slide-1
SLIDE 1

1 SPARC Lab

Shreyas Sen, Ayan Biswas, Priyadarshini Panda, Kaushik Roy ECE, Purdue University Sep 27, 2016

Neuromorphic Computing in CMOS: Digital, Analog or Mixed-Signal ?

slide-2
SLIDE 2

2 SPARC Lab

Neuromorphic – Fundamental Question

Error-Resiliency  Energy-efficiency

  • Approximate Neurons ?
  • Noisy Neurons ?
  • Digital, Analog or Mixed-Signal?

Neuron Architecture

slide-3
SLIDE 3

3 SPARC Lab

Von-Neumann Computing Digital vs. Analog [1]

Digital vs. Analog

[1] Sarpeshkar, Rahul. "Analog versus digital: extrapolating from electronics to neurobiology.“ Neural computation 10.7 (1998): 1601-1638.

slide-4
SLIDE 4

4 SPARC Lab

Neuron Architecture - Digital

Multiplier Adder Digital MAC (n=2) Transistor Count Thresholding

8b: High Static Leakage

slide-5
SLIDE 5

5 SPARC Lab

Neuron Architecture – Mixed-Signal

  • Resistive Load
  • No PMOS load to ensure
  • 𝑺𝒎𝒑𝒃𝒆 ≠ 𝒈(𝑱𝒖𝒑𝒖𝒃𝒎))
  • Large Signal Multiplication

Vout V1 1X 2X 2N-1X N w1 Rload V2 1X 2X 2N-1X N w2 Vn 1X 2X 2N-1X N wn

𝒉𝒏 𝒉𝒏 𝒉𝒏 𝒋𝒄𝒋𝒃𝒕 𝒋𝒄𝒋𝒃𝒕 𝒋𝒄𝒋𝒃𝒕

𝑾𝒑𝒗𝒖 = 𝝉( 𝒙𝒍𝒉𝒏𝑾𝒍

𝒐 𝒍=𝟐

)

slide-6
SLIDE 6

6 SPARC Lab

Comparison: Dig-N vs. MS-N

slide-7
SLIDE 7

7 SPARC Lab

Error Resiliency vs. MS-N Noise

Quantization Noise Thermal Noise

slide-8
SLIDE 8

8 SPARC Lab

Conclusion

Frequency Precision Mixed-Signal Neuron (LS) Digital Neuron

3b 8b 1MHz

slide-9
SLIDE 9

9 SPARC Lab

THANK YOU

slide-10
SLIDE 10

10 SPARC Lab

Dig-N: Noise and BW vs. Power

slide-11
SLIDE 11

11 SPARC Lab

Dig-N: Noise and BW vs. Power

slide-12
SLIDE 12

12 SPARC Lab

28x28 6@24x24 6@12x12 12@8x8 12@4x4 28x28 6@12x12 12@8x8 12@4x4 FCN 10 6@24x24

slide-13
SLIDE 13

13 SPARC Lab

Prec vs error

FCN (I784, 1H 50, 2H 100, O10)

Trained with high precision (16) Scaled down (161) CE as QN increases 5b it can hold accuracy

MNIST CNN MNIST FCN CIFAR CNN

500nA 100nA 10nA 1nA

25.6 28.6 35.4 44.5

500nA 100nA 10nA 1nA 500nA 100nA 10nA 1nA

slide-14
SLIDE 14

14 SPARC Lab

Prec vs error

FCN (I784, 1H 50, 2H 100, O10)

Trained with high precision (16) Scaled down (161) CE as QN increases 5b it can hold accuracy

MNIST FCN

500nA 100nA 10nA 1nA

CIFAR CNN

25.6 28.6 35.4 44.5

500nA 100nA 10nA 1nA

MNIST CNN

500nA 100nA 10nA 1nA

slide-15
SLIDE 15

15 SPARC Lab

MNIST

FCN (I784, 1H 50, 2H 100, O10) CNN

Baseline error is low (good training) Resilient to introduction of thermal noise 50mV(low SNR) diverges more CNN is better trained – Shared weight More neuron, bigger network – more error averaging

Includes retraining for 3b 12%