Just finished a webinar of Kotlin education and Ko...
# science
a
Just finished a webinar of Kotlin education and Kotlin for Science (as always). Slack Conversation
👍 4
f
Watched! I’m waiting for more content on Kotlin for Data Science and Scientific Computing. Which lib has better performance for linear algebra , math, or multik? Has multik some specific application in which overperforms kmath?
a
I will answer questions about performance only if you will explain why do you need performance and what specific application do you have. Talks about performance a usually long and useless. The same about data science. There is no such discipline as data science. There are specific tasks and practices.
f
Experiments with multiple matrix operations, reduce ops and slices through ndarrays. Experiments like these take a lot of time. To shed some light I used Particle Swarm Optimization algorithm for optimize Neural Networks and others Machine learning models Parameters. These were applied to the problem of time series forecasting. there is some component of linear algebra (matrix and vector operations), loops through data, sliding window and lots of parameters combinations parameters and datasets.
I need to implement the Optimizers by hand to test a lot of ideas.