Book Review - Hands on Machine Learning(C1-2)

1 분 소요

All of materials(quotes, images, definitions) are from this book.
It’s all just for self-study.

Chapter 1

incapable of learning incrementally
take a lot of time and computing resources
need to train a new version of the system from scratch on the full dataset(not just the new data, but also the old data)
- then stop the old system and replace it with the new one

train the system incrementally
each learning step is fast and cheap
can learn about new data on the fly, as it arrives
online learning is great for systems that receive data as a continous flow and need to adapt to change rapidly or autonomously
One important parameter of online learning systems is how fast they should adapt to changing data : learning rate
- set high learning rate : then system will rapidly adapt to new data, but it will also tend to quickly forget the old data
- set low learning rate : then system will have more inertia, also be less sensitive to noise in the new data or to sequences of nonrepresentative data points(outliers)
A big challenge with online learning is that if bad data is fed to the system, the system’s performance will gradually decline
To reduce this risk, need to monitor your system closely and promptly switch learning off if you detect a drop in performance

The system learns the examples by heart, then generalizes to new cases by comparing them to the learned examples, using a similarity measure

Build a model of these examples, then use that model to make predictions
training the model : feed algorithm your training examples and it finds the parameters that make the model fit best to your data