7 ways to get started into ML, from easiest to hardest
Machine learning is about teaching computers how to learn from data to make decisions (or predictions). Sitting at the intersection of computer science and statistics, and you may see several buzz words like:
Data Science, Big Data, Artificial Intelligence, Predictive Analytics, Computational Statistics, Data Mining, Supervised Learning, Reinforcement Learning, etc.
So, how to break into machine learning?
1. Watch ML related videos.
So, you want to get started into ML? The good news is that there are countless tutorials all over the internet. Instead of sifting through the sheer amount of content, YouTubers picked the interesting ones, made it concise, and possibly explain in layman terms.
This article introduces seven YouTube channels that aim to uncover the latest and coolest developments in machine learning. For instant, Two Minute Papers videos are entertaining, engaging, and an absolute joy to watch.
Many conferences invite notable speakers and authors to present their papers for those of you who prefer to listen to the industry experts and researchers themselves. Conference on Neural Information Processing Systems (NeurIPS), one of the leading conferences in machine learning, publishes their free access videos.
2. Read about trends and happenings in ML.
At Towards Data Science, writers present written informative articles to a broad audience of readers. Whether you are looking for steps to learn more about data science, get the latest updates on machine learning, or hear pieces of advice on artificial intelligence career, Towards Data Science gives you a learning experience in an all-rounded manner.
Other notable sources to keep you up to date with machine learning are
You can also connect with people who care about ML on LinkedIn. They will naturally share how-tos tutorials and the latest ML developments.
3. Take ML courses.
For starters, the Elements of AI is a series of free online courses suitable for a broad audience to learn what ML is, what can (and can’t) be done with ML, and how to start creating ML models.
AI For Everyone by Andrew Ng is an excellent introduction to ML; it provides a comprehensive overview of what ML is and various concepts, terminology, and methods surrounding ML. This course is mostly non-technical, suitable for anyone, an excellent course for one to start.
For those who want to deep dive, you can start by choosing which specialization you are interested in the Deep Learning Specialization course. Whether it is computer vision or natural language processing, Andrew Ng teaches the key concepts and the mechanics behind how the algorithms work. fast.ai provides free online courses, a practical approach that compliments Andrew Ng’s, by providing hands-on to working on PyTorch.
For those who are interested in working on TensorFlow, Google has the Machine Learning Crash Course with TensorFlow APIs. It is a practical introduction to machine learning, with video lectures, real-world case studies, and hands-on practice exercises for you to follow through.
4. Write about your ML journey.
Regularly consuming information isn’t an ideal way to learn and grow. After all your readings and learnings, the process of writing down your thoughts helps to straighten out your thinking. It is a time and a space to process that information and reflect on what you have learned.
The process of creating content while consuming more content, writing allows you to reinforce your learning. As a data scientist, effective communication is one of the most important non-technical skills to hone.
You can start with writing on Towards Data Science (how to contribute). Our editors help writers to present their ideas by providing feedback. Publishing in Towards Data Science enables you to reach a broader audience by relying on different traffic sources such as Medium’s feed, social media, and newsletters.
Alternatively, you can host your website, full control of the content and the site’s style. I would suggest checking out GitHub Pages. Hosted directly from your GitHub repository, your content will be assessable from the URL with your username like https://username.github.io.
5. Start on ML pet projects.
Now, this is when things are getting more hands-on. Courses help to build foundational skills but working on projects is a great way to learn. You’ll be forced to think critically about the problem and solution all on your own. It builds specific knowledge and skills that can’t be taught; watching someone else doing it is much easier than learning how to do it yourself.
A pet project allows you to explore and apply what you have learned on a deeper level. From a career perspective, this will enable you to build a portfolio and showcase it to potential employers. You’ll know how to ask the right questions, especially how to Google the right questions. It exposes you to a wide range of problems in an end-to-end machine learning project.
Here is an example of a pet project. A web-based project that pulls stock prices from online API and performs predictions using Long Short-Term Memory with the TensorFlow.js framework. View my projects in what I call, Playground.
One of the hardest things when deciding to work on a pet project is finding time to get started. Because most of us have a full-time job, and whatever leftover “free time” there could be used to watch Netflix. But the learning and growth you get from working on a personal project is gold.
6. Compete in Kaggle competitions.
Kaggle can be a great learning tool for beginners, as each competition is self-contained. You don’t need to scope the project and collect those data, which frees you up to focus on solving the problem. Furthermore, the discussions and winner interviews are enlightening. Each competition has its discussion board and debriefs with the winners. You can peek into the thought-processes of more experienced data scientists.
The competition will also force you to tackle every step of the applied machine learning process, including exploratory analysis, data cleaning, feature engineering, and model training. Teaming up can be a great way to collaborate with others who may have more domain knowledge than you do, further expanding your opportunities and learning from others.
7. Work for ML startup.
Joining a startup is an excellent opportunity to learn end-to-end dev ops as a data scientist. There is so much to build, and you will be doing everything, from acquiring and cleaning data, organizing and understanding the data, building and deployment models.
If the sink or swim attitude at a startup doesn’t intimidate you or turn you off to working at a startup, it can be a great way to gain a number of valuable skills quickly. Since you will wear many hats, and you will need to learn things on the fly, you will become better with machine learning and the processes surrounding it to bring it to production.
There you go, 7 ways to get started in machine learning, ranging from easiest to hardest.
- Watch ML related videos
- Read about trends and happenings in ML
- Take ML courses
- Write about your ML journey
- Start on ML pet projects
- Compete in Kaggle competitions
- Work for ML startup
It doesn’t matter which stage you are currently on. Pick the one that suits your learning style, which is comfortable for you — the right balance between the perceived challenges of the task and one’s perceived skills.
You should never quit doing something because it’s hard. Nothing worth having comes easy.
Get started now? Learn about how convolutions work. Or learn more about Transformers and Attention.