Sleep-Like Unsupervised Replay Improves Performance when Data are Limited or Unbalanced
This addresses the problem of data efficiency in machine learning for researchers, though it is incremental as it builds on existing sleep-inspired methods.
The study tackled performance degradation in artificial neural networks with limited or imbalanced data by implementing a sleep-like unsupervised replay phase, resulting in a significant accuracy boost of up to 10% on MNIST and Fashion MNIST datasets when using 0.5-10% of the data.
The performance of artificial neural networks (ANNs) degrades when training data are limited or imbalanced. In contrast, the human brain can learn quickly from just a few examples. Here, we investigated the role of sleep in improving the performance of ANNs trained with limited data on the MNIST and Fashion MNIST datasets. Sleep was implemented as an unsupervised phase with local Hebbian type learning rules. We found a significant boost in accuracy after the sleep phase for models trained with limited data in the range of 0.5-10% of total MNIST or Fashion MNIST datasets. When more than 10% of the total data was used, sleep alone had a slight negative impact on performance, but this was remedied by fine-tuning on the original data. This study sheds light on a potential synaptic weight dynamics strategy employed by the brain during sleep to enhance memory performance when training data are limited or imbalanced.