Practical Deep Learning for Coders with fast.ai & PyTorch Book Chapter wise Summary Points
- This is completely based on Practical Deep Learning Fast.ai course
[Chapter 1] Your Deep Learning Journey
A regression model is one that attempts to predict one or more numeric quantities, such a temperature or a location. Sometimes people use the word regression to refer to a particular kind of model called a linear regression model.
The entire purpose of loss is to define a “measure of performance” that the training system can use to update weights automatically.But a metric is defined for human consumption, so a good metric is one that is easy for you to understand.
Images from a time series dataset using a technique called Gramian Angular Difference Field.
why do we need train,val and test
We use validation loss to kind of of tune the hyper-params. By doing so, we are indirectly using the val data. That’s why we need test data also.
Train data is used to train the model. (Training Process)
Val data is used to improve the model. (Modeling Process)
Test data is used to only evaluate the model. Don’t see it or share, just use it for evaluation.
the automatic training process with backpropagation, the more manual process of trying different hyperparameters between training sessions, and the assessment of our final result.
If your data includes the date and you are building a model to use in the future, you will want to choose a continuous section with the latest dates as your validation set.
[Chapter 2] From Model to Production
- First try to Make and end to end pipeline for the project you want to work.
- Then iteratively modify the parts, like data is not enough or need to use another architecture etc.
- The Drivetrain Approach
- Define your objective like what outcome do you want.
- Define the actions which you can do to get there like what inputs you can control.
- What type of data can help or you can collect.
- Then build a model that you can use to determine the best actions to take to get the best results in terms of your objective.