Transfer learning is the best way to get started with deep learning. In this video, learn what is involved in transfer learning.
- [Instructor] One of the best ways to get started … with deep learning, is to use transfer learning. … Transfer learning is reusing a trained neural network. … This trained neural network would have been trained … on a data set that is similar to the one you're working on. … Many image classification models are trained on ImageNet, … this has 1000 classes of objects and animals. … Now if you're tryna tell the difference … between different types of flowers, … then transfer learning using a model … that was trained on ImageNet would be a very good start. … This is because the edges, colors and textures of objects … that would have been activated … in the first few layers of the model, … will also be relevant with the data set … with different kinds of flowers. … When you're tying to distinguish between flowers, … this will only take place at later layers … of your deep learning model. … The first thing we do in transfer learning … is to remove the top of our pre-train network. … We then replace this with our own custom classifier. …
- Components in the Apache Spark ecosystem
- What is deep learning?
- Using deep learning in Spark
- Working with images in Spark
- Using pre-trained models
- Testing your model
- Deploying models as SQL functions