Dominating Bot Challenges Donkey Kong Country with Simple Code.
Vložit
- čas přidán 25. 10. 2021
- Check out some of your favorite bosses and learn some AI reinforcement learning using "The Brute". Unlock new career opportunities and become data fluent today! Use my link bit.ly/30LjRMp and check out the first chapter of any DataCamp course for FREE!
👊You got something to say to me?? Join our discord!✌️
/ discord
💻My Updated Bot Code.🖱
github.com/ClarityCoders/Gym-...
replit.com/@JacobLower3/Gym-R...
This video was sponsored by DataCamp - Věda a technologie
Unlock new career opportunities and become data fluent today! Use my link bit.ly/30LjRMp and check out the first chapter of any DataCamp course for FREE!
Hey, currently studying computing science in my second year of college, your channel has reignited a passion and sense of enjoyment for computing I thought I had lost for good 👍
That's really an amazing comment thanks for sharing. I love working on these silly projects and teaching people so this means a lot.
Now I wanna see this play Pokemon Gen 1 games. Almost worth giving it a go.
Great video as always!
I was a big Pokemon guy back in the day.
Nice presentation upgrade, feels like I'm at the Clarity Coders TED talk
Thanks for the feedback I was curious on the new format and feel.
@@ClarityCoders I think it's great for explaining concepts for novices like myself, and it's also very helpful to see the practical application and use in the video as well. Keep up the good work!
fascinating job and explanation.. would love if you could make a series on how to build it from start to finish with explanation on each line and what it actually does.
I'll keep that in mind. I always wonder how in-depth people would like me to go.
Great video! What algorithm would you choose for a Match-3 game where the drops from the top always are random? It would be really fun to watch a video of you creating a AI to beat a match-3 game...not candy crush!! ;)
I'll put it on my list and check it out. Thanks for the comment and watching.
Very cool concept of blindly applying min-max, never saw this idea before; I wonder if some type of reward prediction can be used to prune some trees and speedup the process, due to 99% of the time a new node will reward the same as the previous. Offtopic: There is a small typo "episdode".
Great feedback and thoughts! I think that would be a great idea along with adding multiprocessing and some other speed ups. Dang it on the typo doesn't surprise me haha...
Thanks for watching and commenting I really appreciate it.
this reminds me of the old coding train video with the smart rockets
4:37 DataCamp 👍
The 3 second preview CZcams shows for the TV app was just the Rett & Link part.
I was rather confused.
Darn haha....
I've never done anything on Replit, how's the performance vs your "standard" laptop?
I really like it but the performance isn't there for any really heavy AI related tasks. For bots on stuff like discord and reddit it's my new go to host!
In this case though sense this algorithm isn't heavy (No image processing) it's about half the speed of my laptop. Which is a year old gaming laptop so not exactly standard I guess.
It would be interesting to see this algorithm applied to an environment infinitely generated.
Great idea! Thanks for watching and taking the time to comment.
This algorithm only works on deterministic environments unless you mean in a procedurally generated map with a constant seed. This isn't a very smart algorithm, looks a lot like a greedy graph search algorithm + a randomization factor.
@@TeenNewsLive Yup! We cover that a bit in the video as well. Could be infinitely generated as you said as long as it's the same every time. Great comment.
How would the performance of a blind learning algorithm like this compare to a non-blind one, e.g. one that would learn what to do based on what's on the screen at any given moment?
That would be a interesting topic to show I might do that actually!
My immediate thought would be a typical vision CNN say with a DQN would take longer to train and more processing power. However, it would adapt to new scenarios faster. So a long level for example the Image based one would really shine. On a really short level think the rat boss the blind one will outperform for sure.
@@ClarityCoders Interesting, thank you!
Beginner here so doing bots for games was AI or ML and what is the difference between them
AI is a broader term and I get arguments on here all the time but....
AI to me is giving a program a way to interact with a human like interface or a human like way. ML is allows the machine to learn on its own so you don't have to code rules and such. To me all ML is a form of AI but not all AI is ML
please speak bot chess, training pieces and moves
rocket league when? :P
Soon. Very soon. Thanks for watching and commenting!
I like you videos more that code bullet's because he only puts out like 4 per year.
Only way I can beat him haha
TENTH COMMENT!!!!
My man! Thanks for taking the time and watching.
:) You should use the term machine intelligence or child, using the term "it" is rather insulting... This machine intelligence is like your child and will grow up depending on your love and kindness, I want to see your child grow and experience other games... and helping your child experience the world.
I will love my children more in the future. Thanks for watching and commenting.