Neural Network Retraining for Model Serving

04/29/2020
by   Diego Klabjan, et al.
7

We propose incremental (re)training of a neural network model to cope with a continuous flow of new data in inference during model serving. As such, this is a life-long learning process. We address two challenges of life-long retraining: catastrophic forgetting and efficient retraining. If we combine all past and new data it can easily become intractable to retrain the neural network model. On the other hand, if the model is retrained using only new data, it can easily suffer catastrophic forgetting and thus it is paramount to strike the right balance. Moreover, if we retrain all weights of the model every time new data is collected, retraining tends to require too many computing resources. To solve these two issues, we propose a novel retraining model that can select important samples and important weights utilizing multi-armed bandits. To further address forgetting, we propose a new regularization term focusing on synapse and neuron importance. We analyze multiple datasets to document the outcome of the proposed retraining methods. Various experiments demonstrate that our retraining methodologies mitigate the catastrophic forgetting problem while boosting model performance.

READ FULL TEXT
research
01/07/2020

Frosting Weights for Better Continual Training

Training a neural network model can be a lifelong learning process and i...
research
07/31/2019

Overcoming Catastrophic Forgetting by Neuron-level Plasticity Control

To address the issue of catastrophic forgetting in neural networks, we p...
research
05/04/2022

Explain to Not Forget: Defending Against Catastrophic Forgetting with XAI

The ability to continuously process and retain new information like we d...
research
06/16/2023

Catastrophic Forgetting in the Context of Model Updates

A large obstacle to deploying deep learning models in practice is the pr...
research
12/16/2022

MEIL-NeRF: Memory-Efficient Incremental Learning of Neural Radiance Fields

Hinged on the representation power of neural networks, neural radiance f...
research
02/18/2023

On Handling Catastrophic Forgetting for Incremental Learning of Human Physical Activity on the Edge

Human activity recognition (HAR) has been a classic research problem. In...
research
08/11/2022

Empirical investigations on WVA structural issues

In this paper we want to present the results of empirical verification o...

Please sign up or login with your details

Forgot password? Click here to reset