“Machine learning engineering is becoming a real ecosystem
Looking at software engineering as a parallel, we typically see ecosystems “mature” in pretty standard patterns.
A new programming language will emerge with some exciting features, and people will use it for specialized use-cases, research projects, and toys. At this stage, anyone who uses it has to build all their basic utilities from scratch.
Down the line, people within the community develop libraries and projects that abstract common utilities away, until the tooling is capable and stable enough to be used in production.
In this approach, a team takes a pre-trained model and removes/retrains the last layers of the model to focus on a new, related task.
” <— Read the whole thing: “Deep learning isn’t hard anymore – At least, building software with deep learning isn’t”
The above article links to: Transfer Learning - Machine Learning’s Next Frontier
I think it is and it isn’t. It is but as a newbie there are far too many choices. Hard to figure out where to start!