pytorch_connectomics

pytorch_connectomics copied to clipboard

Hello, how do I set up a configuration file for unsupervised learning?
My settings are as follows: DATASET: IMAGE_NAME: train.tif LABEL_NAME: None AUGMENTOR: ADDITIONAL_TARGETS_NAME: None ADDITIONAL_TARGETS_TYPE: None But there is a bug in the program during runtime. The configuration files you give are all examples of supervised learning. Can you give examples of unsupervised or semi-supervised configuration files? And how to write the configuration file is explained in detail in the official document?

hello @liyun-lu, currently the unsupervised and semi-supervised learning functions are under development. we will let you know when they are supported.

Hi @liyun-lu, please check our technical report (https://arxiv.org/abs/2112.05754) for the use case of self-training (a type of semi-supervised learning) in Section 5.1. More functionalities will be provided.