5 ways to get started with machine learning

April 4th, 2019
Introduction to machine learning
Let's read the text

Required reading time: 6 minutes

Kathleen and I are often asked how we approached artificial intelligence and machine learning. That is why we have summarised how we succeeded in entering the market. This collection of links and information is useful for anyone who has discovered their enthusiasm for data science and programming and wants to follow the white rabbit.

1. research with Google Scholar

Our enthusiasm for artificial intelligence (AI) and machine learning began in 2016 when we discovered Pinterest's In image search. It quickly became clear that machine learning (or machine vision) was behind the technology of the visual search engine. Our enthusiasm was so great that we were looking for more information on the web. The inner motivation was an important driver of our knowledge journey. Therefore, I can recommend to choose an area or application of the AI that personally arouses the greatest curiosity. This is then the starting point and the first search word around which your searches can revolve.

An intrinsic motivation is the safest key to success.

We began to sift through Google's catalogue of all scientific papers: Google Scholar. Here we discovered international statistics, patents, current research and studies on visual search, AI or deep learning.

  • Many documents can only be viewed for a fee. Googling the title in the regular search engine often leads directly to the PDF - without registration or costs.
  • Artificial intelligence is a globally researched and relevant topic. Therefore the English search is much more productive than the German search.

At the beginning we ignored the methodological part of the scientific work. This contains complex mathematical formulas or technical details that were not understandable for us at that time. The valuable part for all beginners is the introduction and the summary of the results. Here you will find the essence of research, similar work and thus the next step in the process of continuing education. The references to relevant literature or experts in the field are also valuable. This type of research offers a good introduction to practical applications, the current state of research and leading minds in the field.

Some documents we've dealt with:

If you want to automatically receive the latest and most important research work, I recommend the newsletter of topbots.com. Here interested people regularly get an overview of new papers and works directly into the mailbox.

2. the good old textbook

Those who are not convinced by research on the web can fall back on the classic textbook. There is a varied selection for beginners, advanced and experts. Personally, the book "Coding for dummies" I fell. It contains numerous references to exclusive online trainings that make you familiar with programming in a very entertaining and practical way. Of course, the book also contains a chapter on machine learning.

Books on artificial intelligence

At various events we also obtained recommendations from professors and subsequently acquired the following titles:

These books are still a great help in preparing lectures or working on new aspects of the topic. Of course, there is also literature that ignores the technical details and focuses on a practical reference. These include, for example, "Weapons of Math Destruction" by Cathy O'Neil or the entertaining (and German-language) work "Quality Land" by Marc Uwe Kling.

3. AI further trainings on the web

Since AI is currently thrilling the masses, almost all (elite) universities offer further education on the subject: Harvard, MIT, Oxford (...). These vary in terms of cost (usually between $2,000 and $3,000), time (6 weeks) and content. As a rule, these renowned institutions offer a good overview of the theory of machine learning or artificial intelligence, with few technical details. A cheaper alternative are private online academies such as Udacity, Udemy or Coursera, which provide access to recognised courses (usually between 20 and 80 euros) from Stanford University, for example.

Kathleen and I chose the Massachusetts Institute of Technology because it is a technical college with a clear historical reference to artificial intelligence. In addition, the Training on Business Use Caseswhich was very convenient for us as marketing experts. We got a deep insight into machine learning and got to know application cases. More important was the goal of the course to identify use cases and to develop a (business) strategy for the implementation of AI. This required a low, four-figure investment, which was, however, worthwhile. The course could be implemented part-time.

If you want to learn the practical implementation instead, you can do this for a smaller budget at Codeacdemy.com, for example. There are various options for acquiring knowledge here. One Beginners course costs e.g. 20 Euro per month and starts with the programming language SQL and leads to Python and machine learning. Who is obtain an official certificate needs to invest 200 Euro & 7 weeks time and then received a complete introduction. I am personally a big fan of the courses, as the complexity slowly increases and the sense of achievement per learning step keeps you happy.

In the end, the required skillset is a combination of three things: General understanding of AI and machine learning and basic knowledge of data science as well as a programming language such as Python, R+ or JavaScript.

4. offline courses and studies

We had the fortunate opportunity to attend a practical course at the MIT in parallel to our further training. Code University of Berlin to be able to complete the course. Unfortunately, this offer cannot be booked regularly, but is a special program for companies and their employees. In project groups we got to know the complete implementation process of ML: This started with the installation of Python, led to the definition of a use case and the finding of a database up to the research and application of algorithms. The result was then incorporated into a prototype.

Such comprehensive courses are costly and time-consuming. Support from your own employer can therefore be helpful. Ultimately, companies benefit from employees who can combine the context of their own area of responsibility with ML and thus potentially increase efficiency.

Anyone considering a fully-fledged course of study now has more and more choice. Both local and international universities offer a variety of degrees (Bachelor, Master, MBA) and subject orientations. Anyone who used to have to settle for just one semester of AI in computer science can now devote himself exclusively to intelligent systems. Although there is still a lot of catching up to do here in Germany, almost all faculties are recognising the need and are beginning to change their courses. For example, we are looking forward to Andrew Ng's and Imperial College London's Master's programme in Machine Learning, which will start in autumn 2020.

5. AI networks and workshops

Until then, there are various networks that you can also join as a layman. There, information about workshops, hackathons and conferences is often shared and knowledge is exchanged. One example is the international and inclusive network Women in AI.

A regular exchange with other enthusiasts also ensures that you don't lose touch. Artificial intelligence is developing faster than ever thanks to the great and broad enthusiasm for the subject. Continuing education is therefore a never-ending part of working with AI.

Do you know of any other networks or training opportunities? Then post your recommendation in the comments.

These contributions should please you


  • Reply Michael Petzner April 17th, 2019 at 9:38am

    Hello, Tina,

    Thank you very much for your great contribution!

    I'm really curious how everything will change with the Internet of Things and the AI.

    best regards

    Comment now