Talk:Incremental learning
Add topicLatest comment: 9 years ago by Marko Tscherepanow in topic Incremental vs. online learning
 | This article is rated Start-class on Wikipedia's content assessment scale. It is of interest to the following WikiProjects: | ||||||||||||||||||||
| |||||||||||||||||||||
Incremental vs. online learning
[edit]In the computer science and more accurately in machine learning field, the incremental algorithms has great experimental interesting and there are many papers that refer to them. However, there is a confused between incremental and online learning. From my research the online algorithms can apply incremental learning but it is no necessary each incremental method be online algorithm. Voutmix (talk) 09:30, 15 November 2016
- I totally agree. In order to distinguish between both types of approaches I always defined online learning as learning sample by sample. This may be prone to catastrophic inference or incremental. The incremental fraction of these algorithms is able to keep all (relevant) information learnt before. On the other hand, there are incremental approaches, which are not trained online. They rather use small batches of data for training. Marko Tscherepanow (talk) 09:48, 29 April 2017 (UTC)