- Reinforcement Learning with TensorFlow
- Sayon Dutta
- 82字
- 2021-08-27 18:51:57
Overcoming the limitations of deep learning
These two possible problems can be overcome by:
- Minimizing the use of the sigmoid and tanh activation functions
- Using a momentum-based stochastic gradient descent
- Proper initialization of weights and biases, such as xavier initialization
- Regularization (add regularization loss along with data loss and minimize that)
For more detail, along with mathematical representations of the vanishing and exploding gradient, you can read this article: Intelligent Signals : Unstable Deep Learning. Why and How to solve them ?
推薦閱讀
- Spark編程基礎(Scala版)
- Learning Apache Cassandra(Second Edition)
- LAMP網站開發黃金組合Linux+Apache+MySQL+PHP
- Salesforce for Beginners
- 空間機器人智能感知技術
- 計算機組裝與維修實訓
- Windows 7故障與技巧200例
- 百度智能小程序:AI賦能新機遇
- Microsoft 365 Mobility and Security:Exam Guide MS-101
- 工業機器人與自控系統的集成應用
- 傳感器與檢測技術
- Mastering Windows Group Policy
- Hyper-V Security
- 實戰GAN:TensorFlow與Keras生成對抗網絡構建
- 深度學習