Improving Network Training on Resource-Constrained Devices via Habituation Normalization
As a technique for accelerating and stabilizing training, the batch normalization (BN) is widely used in deep learning. However, BN cannot effectively estimate the mean and the variance of samples when training/fine-tuning with small batches of data on resource-constrained devices. It will lead to a...
Main Authors: | Huixia Lai, Lulu Zhang, Shi Zhang |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2022-12-01
|
Series: | Sensors |
Subjects: | |
Online Access: | https://www.mdpi.com/1424-8220/22/24/9940 |
Similar Items
-
Stressors Length and the Habituation Effect—An EEG Study
by: Izabela Rejer, et al.
Published: (2022-09-01) -
Dynamic Causal Modelling of the Reduced Habituation to Painful Stimuli in Migraine: An EEG Study
by: Iege Bassez, et al.
Published: (2020-10-01) -
Self-Learning Pipeline for Low-Energy Resource-Constrained Devices
by: Fouad Sakr, et al.
Published: (2021-10-01) -
Habituation of the stress response multiplex to repeated cold pressor exposure
by: Tom Bullock, et al.
Published: (2023-01-01) -
Longitudinal fNIRS and EEG metrics of habituation and novelty detection are correlated in 1–18-month-old infants
by: Laura Katus, et al.
Published: (2023-07-01)