6. März 2018 5 min to read

Creativity, Innovation, Intersections and Deep-Learning

Category : Books, Deep Learning, Natural Language Processing, Use-Cases

Namaste! I am writing this article on a very silent evening after a very productive day. You know… Doing things. Learning things. Teaching things. Gathering knowledge and applying it. And what is the result? A word-of-the day. You want to know what it is? It is „bag-of-notes“. I will talk about that later. But let me give you a hint: It has to do with music and Deep Learning. It is an example of intersections in different fields.

Creativity… It’s a very, very random dancer.

Recently, I read a book that moved me a lot. The Medici Effect by Frans Johansson is a blast to read! What is it about? Creativity and innovation. Especially in our times this is very important. There are so many established companies around who struggle with both. It is the age of Digitalization. The age in which many institutions move towards more and more Information Technology. It is the years of the next industrial revolution, which is all about IT and AI.

You know… I am a very creative individual. I tend to get inspired by many things. You can see this in all my articles. One day I write about applying Deep Learning to the weird fiction works of H.P. Lovecraft and on the other day about simulating diseases with Cellular Automata and generating data for Deep Learning.

I know that creativity is a matter of coincidences. It is random. It is noise. Everyday we are flooded with tens if not hundreds of ideas. On public transport. While mowing the lawn. In the shower. Everywhere. But how can we make such ideas successful? How can we use creativity to fuel innovation and shape the face of our planet? Where does true innovation come from in these times? In times that are changing rapidly?

Innovation happens where differences intersect.

The Medici Effect states that innovation happens where different disciplines, cultures and industries intersect. What a statement! Read the state again if you like to let it sink in. And how true it is… It is diversity that fuels innovation. Just take two people from absolutely different backgrounds and let them talk for an hour over a nice cup of coffee. It is highly probable that you will end up with way more ideas than when doing the same with two people from the one and the same field for a week.

In his book Johansson shows that many great things in our lives are rooted in diversity. There is a lots of examples for great business that came into being after finding a fruitful intersection. Remember Virgin the record label? After quite some success making and selling music the company decided to enter a completely different field. The people managed to create a quite successful airline. Yes, flying planes. And how did they do that? They took their understanding of excellent customer care from the music business to airlines. Instant success.

Intersecting Deep Learning with… Everything.

My favorite technology is Deep Learning. I see diversities intersecting. Galore! Every day! An example… Last year I gave a talk at Würzburg University. Invited by Prof. Fotis Jannidis, chair of Literary Computing. I did not expect that my interests had so many intersections with humanities. Truth be told. I am a computer scientist by education. In Germany, computer science is a natural science. Literary computing on the other hand is a humanities. It deals mainly with computer linguistics. Processing texts with the support of computers. Yet, I was there discussing Deep Learning and its applications in Natural Language Processing. This created lots of ideas. That was an intersection of different fields!

And another story… My partner in life also comes from the non-natural-sciences. She has a Master’s degree in India Studies and Digital Humanities. Her focus is culture. For me that is data. For her, the computer is just a tool. For me, it is a life-changer. Every day we see the differences in our perspectives. But also every day, we talk about the same things. It is not rare that we do brainstorming on the intersection. Often we come up with many challenging insights and promising ideas.

If I would tell you that the intersection between music and Artificial Intelligence is a very exciting one… Would you be surprised?

I promises you the word. Bag-of-notes.

One of my most-favorite intersections is Deep Learning and music. Some of my first programs were simple and crude music generators. Whoa… That was back in the nineties. Somewhere I got a list that mapped musical notes to frequencies. And I found some code that would play these notes on the PC-Speaker. Programmed in QBasic. Play one note after another and you get a melody. Easy as that!

Now it is the year 2018. One of my focusses is sound analysis and sound synthesis. How to get information out and how to generate new ones. There is one thing that I have to admit now. This is something that is very important for Deep-Learning. The Neural Networks that I train have absolutely no opinion about the data I feed them. They are all joyfully agnostic and do not care if they train on natural language or on sounds. Or on images… Or on weather data… Or…

Time to raise the curtain. What is the bag-of-notes thingy? It is encoding notes, intervals and chords in the same way as texts in Natural Language Processing. Remember… In Natural Language Processing there are many ways how to encode texts. Using strings is straightforward, but way too inefficient. One-hot-encodings map words to 0-1-vectors. They are easy to understand but they waste a lot of memory. Bag-of-words map sentences to 0-1-vectors. That is a little better, but the ordering of the words goes out of the window. Word-embeddings are great. These are dense vector of a fixed length. And they carry semantics.

Why am I talking about encoding words and sentences? Well.. Again… The Neural Network does not care, if you feed it words and sentences. Or notes and chords. You see what I am getting at? Yes… You can apply techniques from Natural Language Processing to Sound Processing and vice versa! And who is currently writing about that? A computer scientist standing at the very intersection of all those mentioned fields… This is a great time for innovation!

Final words… At least for today.

It all started with the book The Medici Effect. It is a good read. And its insight are marvelous. You cannot fuel innovation with teams that are from the exact seem field. You need diversity. And we are now in the age of diversities. Many fields and cultures developed independently. And now they are converging. Great times! And what is your role?

Stay in touch.

I hope you liked the article. Why not stay in touch? You will find me at LinkedIn, XING and Facebook. Please add me if you like and feel free to like, comment and share my humble contributions to the world of AI. Thank you! I am looking forward to talking to you!

Leave a comment