Building dataset – Deep Learning in Halite AI competition p.3

21 thoughts on “Building dataset – Deep Learning in Halite AI competition p.3”

1. Hello Sentdex, any plans for videos related to tensorflow on nodejs? It seems that nodejs is poised to become a solid competition for python. Just think about, js is a lingua franca of the dev world. One language for front and backend. To prove it I have deployed a v.simple serverless mobile app using MobileNet model with tfjs-node to predict objects from an image. Images are uploaded to cloud with cloudinary.com. Then serverless ibmcloud functions deployed with docker are invoked for the image analysis with node tensorflowjs.
   https://github.com/bartosz-paternoga/MobileNet_tfjs-node_SLS
   demo https://bartosz-paternoga.github.io/mobilenet-tfjs-node-sls
   Could be great if guys like you promote machine learning with tfjs, and not necessarily on the browser alone but on the server side with nodejs as well. Hope to see some js on you channel! cheers

2. See you soon bud

3. One more thing I need your discord adds

4. I meant addr

5. I guess it's better to see how much has been collected by the ship and not how much has been desposited to your shipyard since moving away shouldn't be punished due to random movement.

6. For the beginning part you can just query from constants.MAX_TURNS to find out how many turns there are in the game you're playing

7. When running two instances of the same python program, there's no risk of the variables getting crossed?

8. really loving these series man! thank you!

9. hi i need help sentdex , about keystroke dynamic can i have your email to take about it it importan to me

10. I know this hasn't really to do with the video, but I can't find a solution to my problem:
    I'm trying to make an RNN that also has Conv2D at the start in Keras.
    I used adam optimizer and everything but the LSTM cells is TimeDistributed. It should track the velocity of a dot, but the loss won't go down.
    The problem is a bit to big to explain in full detail here, so if you would like to help me on this, I will explain it better. Thanks

11. Tip: zoom in the recording or increase the font size. Because text is so small for someone watching on smartphone like me.

12. around 1 hour and 30 minutes into this machine learning tutorial i find that i wasted my time and could have done much better with a basic 30 minute video where basic replay data is parsed and then some statistics are piled up to adjust a training value
    after 1 hour 30 minutes this machine learning process is still at the setup stage.

13. Produced 40Go of file data over night lol. Ready for next video ^^

14. You have made so many great AI videos. I just got Anki Cozmo and want to learn how to use Tensorflow with the cozmo SDK so that AI and Neural Networks can be merged with real-life products and robots. Hope you can help.

15. On Linux a turn-limit under 400 does not seem to work

16. Next video maybe face recognition using deeplearning and autoencoder

17. Santdex How can i do prediction personality analysis of multiple CV and then Rank them using python and the CV is stored in Database what techniques should i use Please Guide me
    must reply

18. I would recommend using the following command to make lots of training data:
    
    halite.exe –no-replay –no-logs –results-as-json –turn-limit {MAX_TURNS} -vvv –width {map_size} –height {map_size} "python sentdebot.py" "python sentdebot.py"

    This allows you to not print the replay or error log files. Plus the cmd window game console output isn't text it's the numeric json. This is the best way I found to create training data.

19. Why does he keep pretending to end the video? Is this like playing the YouTube algorithm metagame?

20. When will the code be available on pythonprogramming.net?

21. Hi my question is all dataset can be use in python with tensorflow or not .

Comments are closed.