My father always say
“To explain even most complex thing in simple,easy way is Knowledge.”
In this post , I will be sharing my views on myths that beginners come across while indulging in Deep Learning.
Deep Learning is not complete AI.It is a subfield of Machine Learning which based on biological aspect of brain to achieve or perform a task.
We don’t need to have Ph.D in order to work on it .Although it is complete necessary to have some knowledge on differentiation,mean,standard deviation,distribution ,etc.
- There is difference b/w the terms AI(Artificial Intelligence),ML(Machine learning),DL(Deep learning).Don’t be confuse as some time people use them interchangeable.
- I will encourage you to have some command on numpy and matplotlib or any visuilzation library ,it will be useful if learning from scratch or want to understand things from core.
- Basics will always remain important no matter what others say.Ex.- back-propagation,gradient descent,optimization technique and weight-bias trade off. These are concepts which we you need to have clear understanding.
- Using Library like Tensorflow,Keras,Pytorch is good but also try to read their documentation.It contain useful information
- Don’t be at Rush to execute each model or simply use any code,take your time ,understand and ask from others on opinion on your work.
- Try to read Research paper and try to visualize its perception,slowly but regular reading and you will be able to understand it.
- We don’t need to train model from scratch, we can train them on top of pre-trained model also.
Reading Research paper will be advantageous if you want Deep Learning as carrer path. Ex- you can also go through
Two Minute Paper ( Károly Zsolnai-Fehér)from Youtube
8.Any Deep learning Model follows these steps:
Evaluate and check accuracy
9.There are many good resources to learn and understand about Deep learning.Ex-
CS231n of Standford,
Deeplearning.ai from Andrew Ng,
fast.ai from Jeremy Howard.
Note-All courses are good,free and provide platform for like minded peoples to interact.
Source: Deep Learning on Medium