Kotlin strives for Deep Learning
•
0 likes•651 views
AI is New Electricity and Deep Learning is one of key enablers for this, it breaks known limits of possible and disrupts vast areas of our modern life. From dev standpoint it sounds science-heavy and requires PhD in fact its not. Here I’m going to explain why in theory and practice with Kotlin.
Kotlin strives for Deep Learning
- 1. Kotlin strives for Deep Learning
- 2. ● proud father ● SA in EPAM Systems ● Java is my primary programming language ● infected by AI disrupting power ● passionate about agile, clean code and devops
- 3. Agenda ● Should I care? ● First things first ● Let’s code a bit ● Takeaways ● Q&A
- 8. What is DL?
- 9. DL is subset of ML and AI
- 13. ● Compute ● Data ● Algorithms ● Infrastructure Recent Breakthroughs
- 14. Stop! Where is Kotlin?!.
- 21. Propagations
- 22. Is this enough? ● Algorithms ● Data Handling ● Distributed Computing ● GPU Computing
- 23. What’s Next?
- 24. Komputation ● NN framework for JVM written in Kotlin ● Supports CUDA ● Provides CNN, RNN implementations ● Built-in activation functions like ReLU, Sigmoid, Softmax, Tanh ● Built-in loss functions and optimizers
- 25. KotlinNLP ● Set of libraries for NLP written in Kotlin ● Provides support for ○ tokenization ○ categorization ○ parsing and entities recognition
- 26. References ● Koma http://koma.kyonifer.com/ ● Komputation https://github.com/aisummary/komputation ● Kotlin Data Science Resources https://goo.gl/MDLRHK ● Can I do AI? https://goo.gl/16BdY4 ● Source code https://github.com/webdizz/kotlin-strives-for-dl
- 27. Takeaways ● Disrupt yourself with AI skills to not be disrupted ● AI is not a rocket science you can do it ● Today it’s really easy to learn something new ● It’s still unbelievable what's possible with AI