Hierarchical recurrent encoder-decoder model(HRED) that aims to capture hierarchical structure of sequential data tends to fail because model is encouraged to capture only local structure and LSTM often has vanishing gradient effect.
Reinforcement learning with sparse reward often suffer from finding rewards.
This paper wanted to catch non-linear dynamics of the object in video.
In many enviroments of RL, rewards tend to be delayed from the actions taken. This paper proved that delayed reward exponentially increase the time of conversion in TD, and exponentially increase the variance in MC estimates.
평점: 4.5
데이터의 힘: 아픔이 길이 되려면을 읽고\
Most of deep directed latent variable models including VAE try to maximize the marginal likelihood by maximizing the Evidence Lower Bound(ELBO). However, marginal likelihood is not sufficient to represent the performance of the model.
in Studies on Deep Learning, Deep Learning
Traditional explanation for generalization of a machine learning model was primarily concerned with tradeoff between model capacity and overfitting. If capacity of a model is too large, you must contrain the capacity to prevent overfitting. Choosing appropriate level of capacity of model has been seen as a key to generalized performance.
Rejoice in what you learn and spray it!