Around thirty people gathered at Gneisenaustraße in the evening. It was my pleasure to talk about the Deep Learning experiences I gathered in my 1.5 years at Porsche Digital Lab Berlin. My experiences about Deep Learning, corporate Digitalization, and beyond.
A mantra and a clarification about witchcraft…
I made a couple of important points. There is a mantra I often say. Sometimes I do not like to repeat myself, but this mantra is worth the air: “When it comes to Deep Learning, it is time-to-market. Everyone can get started and train their first Neural Networks in no time. Everyone has access to hardware, software, papers, datasets and best-practices. All you need is to set aside some time and dive headfirst into Deep Learning!” Well, what a mantra!
And I also made explicit that Deep Learning is neither witchcraft nor a blackbox. The blackbox-idea is a no-truth. But it is propagated by way too many people. Nothing could be farther from the truth. Of course, a Deep Learning expert needs some experiences to come up with a great data preprocessing, to create a good Neural Network architecture and to tweak the hyperparameters properly. But at the end of the day, it is all about running into overfitting and then compensating it. This is the whole story!
The giant only looks like one.
You know, there is not so many different types of layers, activation functions, metrics and optimizers. The whole design space of Deep Learning architecture is not so huge. And believe me, even some of the most complex problems can be solved with very simple architectures. You could get quite far with fully connected layers – the most straightforward design of neural networks.
It is always a good starting point to begin with fully connected layers. For example for time series analysis, you could start right with that. Only if the accuracy turns out to be suboptimal you could try a 1D Convolutional Neural Network. And if that fails go for a proper Recurrent Neural Network.
There is no magic about Deep Learning. The message I brought across is that there are some best practices, but they usually do not fill a whole book.
Würzburg has a huge interest in Deep Learning and a growing scene.
I spent a lot of time in Babylon-Berlin. For me it is great to see what is going on in Würzburg. I am amazed. There is a huge Digitalization scene. So many great companies working on so many bleeding edge projects. Design Thinking, IoT, Chatbots, the Cloud… You name it!
It is good to be here. Optimism is my main feeling nowadays. And I am confident that I will have lots of fun here in Bavaria!
Do you want to know more?
Sign up for the Würzburg Deep Learning Meetup Group for updated information and dates. And if you like attend our next meetup where Mayflower’s Alexander Oldemeier will talk about time-series analysis with TensorFlow. Such an exciting topic!
And a final word… I am very new to Würzburg and there is one thing I would love to do: Working with you on great Deep Learning project! Please contact me if you are interested! firstname.lastname@example.org
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!
If you want to become a part of my mission of spreading Artificial Intelligence globally, feel free to become one of my Patrons. Become a Patron!
A quick about me. I am a computer scientist with a love for art, music and yoga. I am a Artificial Intelligence expert with a focus on Deep Learning. As a freelancer I offer training, mentoring and prototyping. If you are interested in working with me, let me know. My email-address is email@example.com - I am looking forward to talking to you!