-
https://github.com/facebookresearch/ResNeXt
-
https://arxiv.org/abs/1804.07612
-
## 論文リンク
https://arxiv.org/abs/1703.00810
## 公開日（yyyy/mm/dd）
2017/03/02
## 概要
Information Bottleneck (IB) principle を DNN に適用するというアイデアを提唱した https://github.com/yoheikikuta/paper-reading/issues…
-
I would like to ask about the upper and lower bounds of the obs space in `BaseRLAviary.py`, I. noticed that the bounds are - and + infinity, does not that make the state space very huge to be explored…
-
-
### Description
I had a deep look into the coolify containers and found some things I believe could be improved (hope I can help like this as I am at the moment not ready to start code contributing)
…
-
Training a deep model usually requires a large GPU memory, which may not be feasible in clinical applications. Therefore, the aim of this issue is to concentrate on possible approaches for reducing me…
-
- https://arxiv.org/abs/2105.04857
- 2021
我々は、学習した深い特徴表現に疎な線形モデルを当てはめることで、よりデバッグしやすいニューラルネットワークを構築できることを示す。
これらのネットワークは高い精度を維持しつつ、人間による解釈にも適していることを、数値実験と人体実験によって定量的に示します。
さらに、結果として得られるスパースな説明が、視…
e4exp updated
3 years ago
-
本文提供了利用frame-level 信息的两种思路：feature-pooling 以及 recurrent neutral networks，前者没有利用temporal 信息，后者则利用了temporal 信息。
（3d-convolution尝试的必要不大，Karpathy 指出这种结构just marginally better than single frame baseline.）…
-
## Addition of an AI Vision-Based Solution for Road Condition Monitoring
The solution aims to automate the detection and classification of road surface defects (such as potholes, cracks, etc.) usin…