Learning AI

Artificial intelligence (AI) enters our lives in many different ways. AI makes it possible for machines to learn from experience, adjust to new inputs and perform human-like tasks. Most AI examples that you hear about today – from chess-playing computers to self-driving cars – rely heavily on deep learning. Deep learning is a branch of machine learning utilizing giant neural networks and massive data sets. Using these technologies, computers can be trained to accomplish specific tasks by processing large amounts of data and recognizing patterns in the data.

Artificial intelligence is growing exponentially in all the major sectors, including health, social media analysis, self-driving cars, language processing and others. The AlphaGo victory is just one of the signs of amazing things to happen. The understanding of artificial intelligence opens lots of opportunities.

If you’re considering working in AI as a data scientist or machine learning engineer but need to find a good starting point, here are a few things to consider in your learning journey:

  1. Get your mathematics strong. You should have some appreciation of the mathematical underpinnings, especially linear algebra and calculus. Specifically, you’ll need to be comfortable with matrix multiplication and partial derivatives.
  2. To get a development role on an AI team, be sure to have at least one to two years of software development and machine learning experience under your belt. This can include building your own projects or working at a company driving key projects such as image or text classification. A great place to look for machine learning projects is arXiv where researchers often publish their papers. You can sharpen your skills by implementing models and systems from papers that capture your interest.
  3. Learn Python. This handy programming language is the tool of choice for most machine learning engineers and data scientists.  Python’s syntax is relatively easy to pick up and it has a vibrant and helpful community. The language also has excellent documentation and tons of training resources. With tools such as Jupyter notebooks and libraries like Numpy and Pandas, Python has become the first choice for developing machine learning and deep learning applications.   Outside of machine learning, Python is useful for developing websites, videos games, and more. Udacity can get you coding in Python and building your first neural network in just three months!
  4. Learn mainstream deep learning libraries like TensorFlow or PyTorch. Most deep learning systems are built in either TensorFlow or PyTorch, Python frameworks that provide APIs for defining and training deep learning models. You’ll want to be experienced with at least one of these frameworks as most AI teams are using them for research or product development. You should also consider joining the machine learning community.  

The biggest area of AI research today seeks to enable computers to make inferences from complex data. Techniques to do this are termed machine learning (ML).  AI and ML are large and rapidly-developing fields. While it’s impossible to capture their full potential in a this blog post, we’d like to invite you to Mat’s workshop on Natural Language Processing at the Global AI Conference on January 24th from 2pm – 6pm PST in Santa Clara, California.

Author: Mat Leonard

Mat Leonard is Product Lead for Udacity's School of Artificial Intelligence. He is a former physicist, research neuroscientist, and data scientist. He did his PhD and Postdoctoral Fellowship at the University of California, Berkeley.