The Law of Parsimony in Gradient Descent for Learning Deep Linear Networks.
Can YarasPeng WangWei HuZhihui ZhuLaura BalzanoQing QuPublished in: CoRR (2023)
Keyphrases
- learning tasks
- learning algorithm
- connectionist networks
- learning process
- online learning
- cost function
- unsupervised learning
- learning systems
- deep architectures
- data sets
- recurrent networks
- deep learning
- learning rules
- learning community
- incremental learning
- complex networks
- pairwise
- objective function
- reinforcement learning
- case study
- social networks
- artificial intelligence
- machine learning