My personal experience (as hobbist): if you share my same background, i.e a degree in Computer Science, we'll, i'd say that many of concepts you may have learned in the late 90's still apply, but at the same time, Deep learning really gone very very far.
Using Tensorflow required me to study from scratch a bunch of concepts, and still now I can't say I master them. At the very end I preferred to use Keras (with tensorflow as backend), much more simpler, and no less powerfull; moreover, Keras hide a lot of the complexity of dealing with tensors,
with gradient descent, with learning rate and so on.
I'd say that AI background isn't strictly required to work with tensorflow; instead, a solid background in calculus and algebra may help a lot.
posted 1 month ago
Thank you. I shall have to remember my algebra from a very long time ago
Thank you for the question. It is a tough one. It will quite depend on what exactly you have forgotten.
If you only need a quick refresher on ML you read a few blog posts / watch few youtube videos and get upto speed.
If you think you need a more grounded walkthrough, I'd suggest reading a book/doing a course focused on Machine Learning. Unfortunately, the machine learning books I've referred during my degree has been quite mathematical. If you don't mind that Bishop's Pattern Recognition and Machine Learning might work for you. If not, unfortunately you'd have to see for yourself as I haven't been in this situation. Meanwhile familiarize yourself with Python and popular tools like Pandas/Numpy/Matplotlib/Sklearn.
Once you are familiar with the popular modelling methods, Python and its popular libraries, you can go for "TensorFlow 2 in Action". Unfortunately, the book assumes moderate knowledge of Python and these libraries. You don't need to be a master. But should be comfortable using them. Then in the book, you will have the required guidance for the following steps in the work flow in a ML project ( primarily from a deep learning perspective).
+ EDA - Exploratory data analysis
+ Data cleaning
+ Feature engineering (typically not used for deep learning. So this will not be covered in this book)
+ Deep network mdoels
+ Model evaluation
PhD | Senior Data Scientist | AI/ML Educator
it looks as a fantastic topic, and I M really enjoying the structure of the book, and the introduction available on line.
A small remark: for what concern linear algebra foundations there is a book called math from Programmers from the excellent Manning editions as well, and it covers the basic of linear algebra, could be a good introduction if a bit rust with vectors and matrix, even because gives some math example in python and slightly introduce data science, anyway all the spotlight on this fantastic book from Thushan and his exciting chapters
We coders are all geniuses. Sometimes misunderstood, others humble, introvert or without focus, but how much light abide in our curiosity?(me)
Power corrupts. Absolute power xxxxxxxxxxxxxxxx is kinda neat.
Building a Better World in your Backyard by Paul Wheaton and Shawn Klassen-Koop