Presented by:

Jadga Hügle
Jadga works on Snap! and loves to create fun projects that might also be helpful for students and teachers. Her favourite Snap! block is map <3

Uwe Lorenz

Jens Mönig
No materials for the event yet, sorry!
Neural networks and deep learning are an important part of current artificial intelligence. Therefore, we want people to understand what's happening under the hood. We've used a single ‘layer’ sprite to create a classic Rosenblatt perceptron. Duplicate the layer sprite multiple times and customise the receivers in the transmission blocks to create deep neural networks. Use the setup script to customise the topology and learning rate of the network. Experiment with exciting data sets, investigate the effects of hyperparameters and observe the data stream processing of the deep neural network. Watch how the network learns by processing the error stream during backpropagation.
- Duration:
- 15 min
- Room:
- Plenary Room
- Conference:
- Snap!Con 2025
- Type:
- Talk
- Difficulty:
- Medium