Journey Toward Artificial Intelligence
A.I, Machine Learning, Deep Learning: what are those terms?
If you are not quite sure about those terms, you are at the right place.
I have compiled the articles and online courses that you can follow this short journey to become a data scientist. You will become familiar with an A.I. area and perhaps become a data scientist at the end of this journey.
The following are my recommendation articles and courses.
1. Understand the differences of the AI, Machine Learning, and Deep Learning by reading the following article so that you understand those terms.
A Simple Way to Understand Machine Learning vs Deep Learning
2. Take the FREE online courses from the fast.ai provided by Prof. Jeremy Howard and Prof. Rachel Thomas. There are two parts to understand the Deep Learning by coding. You will learn how to set up the GPU on Amazon cloud and run your first Deep Learning classification before you deep dive into how it works. I recommend the fast.ai because I like its teaching philosophy. You will learn and familiar with the important algorithms on the way by showing you how it works first before diving into the actual algorithm by coding.
Once you complete the part I, you will feel confident in this journey. You can take the part II of this course as well to gain more insights into the Deep Learning. Part II will also give you the idea how to read the research papers related to A.I from the arXiv.org hosted by Cornell University for example.
3. Learning the TensorFlow library. Google provides the TensorFlow library for free so that you can run your Machine Learning and Deep Learning on your own machine. It has very intuitive tutorials that you can follow step-by-step. The Getting Started guide is the best place for you to start to ensure that you set up your own machine for its library properly. Next, you can follow the Tutorials to see how it works by doing it. In addition, TensorFlow comes with the built-in Estimators to simplify the Machine Learning programs. For example, if you want to run the DNN classifier algorithm, you can use the tf.estimator.DNNClassifier API as the example code below.
As you can see how easy it is to run the DNN by using the TensorFlow library. You don't need to write the DNN algorithm from scratch.
estimator = DNNClassifier( feature_columns=[categorical_feature_a_emb, categorical_feature_b_emb], hidden_units=[1024, 512, 256])
4. At this point, you have become the Machine Learning and Deep Learning developer so congratulations. I would recommend you to sign-up for a free account on Kaggle so that you can run your Machine Learning and Deep Learning projects for FREE. You also can learn how other people solving the actual problems. Kaggle is the Google's subsidiary company. You can share your own dataset on Kaggle. Kaggle hosts the competitions thus if you are interested, you can submit your model to compete with other developers. You may win the big prize!
5. You will need to work on the data manipulation when you work on your model. I would recommend the Pandas library which comes very handy to load CSV file for running your models and predictions. Jeff Delaney's blog, "19 Essential Snippets in Pandas", is very good. You can take a quick look at his's blog to get an idea what Pandas can do.
6. Finally, if you feel like you want to fully understand the Machine Learning deeper, you can take the Machine Learning course by Prof. Andrew Ng at Coursera. You will learn about the machine learning in detail, for example, you will learn how to write the cost function and Sigmoid function without using any library.
I hope you enjoy this journey.