all AI news
Consistency Training with Virtual Adversarial Discrete Perturbation. (arXiv:2104.07284v2 [cs.CL] UPDATED)
April 29, 2022, 1:11 a.m. | Jungsoo Park, Gyuwan Kim, Jaewoo Kang
cs.CL updates on arXiv.org arxiv.org
Consistency training regularizes a model by enforcing predictions of original
and perturbed inputs to be similar. Previous studies have proposed various
augmentation methods for the perturbation but are limited in that they are
agnostic to the training model. Thus, the perturbed samples may not aid in
regularization due to their ease of classification from the model. In this
context, we propose an augmentation method of adding a discrete noise that
would incur the highest divergence between predictions. This virtual
adversarial …
More from arxiv.org / cs.CL updates on arXiv.org
Jobs in AI, ML, Big Data
Data Engineer
@ Bosch Group | San Luis Potosí, Mexico
DATA Engineer (H/F)
@ Renault Group | FR REN RSAS - Le Plessis-Robinson (Siège)
Advisor, Data engineering
@ Desjardins | 1, Complexe Desjardins, Montréal
Data Engineer Intern
@ Getinge | Wayne, NJ, US
Software Engineer III- Java / Python / Pyspark / ETL
@ JPMorgan Chase & Co. | Jersey City, NJ, United States
Lead Data Engineer (Azure/AWS)
@ Telstra | Telstra ICC Bengaluru