{"author_link":"\/users\/orcifried","author_name":"OrciFried","author_uid":"1044013","comments":[{"author_name":"sriram070","author_uid":"1212795","time":"2020-10-05T03:33:13Z","epoch":1601868793,"modified":1601868793,"text":"Great concept and execution. At style was also amazing. But a lot of the time the AI agents would go of the screen and then nothing would happen","likes":1,"format":"md"},{"author_name":"Ividusk","author_uid":"1213567","time":"2020-10-05T03:39:40Z","epoch":1601869180,"modified":1601869238,"text":"Graphics and music are really nice! Perfect for the game.\n\nThis idea is fantastic - however after a few seconds, all the agents disappear on me. I'm not sure if I'm doing something wrong or if it's a bug, but it's a shame because what I did see was really cool!","likes":1,"format":"md"},{"author_name":"OrciFried","author_uid":"1044013","time":"2020-10-05T05:31:24Z","epoch":1601875884,"modified":1601875884,"text":"That is definitely not supposed to happen :sweat_smile: But thank you :blush: ","likes":1,"format":"md"},{"author_name":"FabDynamic","author_uid":"1078813","time":"2020-10-06T21:30:35Z","epoch":1602019835,"modified":1602019942,"text":"Wow the visual aesthetic on this game is through the roof retro-amazing.  And I can't believe that its about training AI!  I tried to get my best agent to run to the end but it never showed up (probably the same problem the other people are experiencing).  You probably know this already but if you have time you can fix bugs and re upload them and its still a valid submission to the compo.  Give me a buzz here if you have any fixes for us and thanks for making this!\n\nUpdate: Also the source link is broken.  I was thinking about grabbing a copy and seeing if I could figure out what was wrong :) Let me know if I can help.","likes":1,"format":"md"},{"author_name":"OrciFried","author_uid":"1044013","time":"2020-10-11T08:44:47Z","epoch":1602405887,"modified":1602405887,"text":"@fabdynamic that is absolutely wonderful, thank you! I haven't had time the past week but I'll try to update both the game and the source code tonight so if you want, it'll be there \u2764\ufe0f","likes":1,"format":"md"}],"format":"md","images":["ld47\/222992-74d07681a5122ea91971d2b2c1ad44ca.png"],"links":[{"url":"https:\/\/esbenkc.itch.io\/loop-to-learn","text":"Youtube"},{"url":"https:\/\/github.com\/esbenkc\/LDJAM47","text":"Youtube"}],"metadata":{"g_key":"65350","g_author":"1044013","g_event":"LD47","g_eventkey":"76","g_subevent":"COMPO","g_urlkey":"282700","g_title":"The Learning Loop","g_status":"PVT1","g_place":"99999","g_commentcount":"5","g_site2_node_id":"222992","g_hide":"N","g_has_icon":"Y","g_rqueue":"0","g_random":"0"},"nds":{"n_key":"222992","n_urlkey":"282700","n_parent":"212256","n_path":"\/events\/ludum-dare\/47\/the-learning-loop","n_slug":"the-learning-loop","n_type":"item","n_subtype":"game","n_subsubtype":"compo","n_author":"44013","n_created":"1601702632","n_modified":"1601878804","n_version":"672653","n_status":"COMMIT"},"node":{"id":222992,"parent":212256,"superparent":9,"author":44013,"type":"item","subtype":"game","subsubtype":"compo","published":"2020-10-04T21:18:54Z","created":"2020-10-03T05:23:52Z","modified":"2020-10-05T06:20:04Z","version":672653,"slug":"the-learning-loop","name":"The Learning Loop","body":"![title.gif](\/\/\/raw\/deb\/a\/z\/358f4.gif)\n\n## A Game of Reinforcement Learning\n\n[Play it here! :heart:](https:\/\/esbenkc.itch.io\/loop-to-learn)\n\nYou are an engineer tasked with optimizing artificial intelligent agents in the completion of spatial navigation tasks to satellites. You have control over the time scale, which aspects of the environment triggers rewards in the agents, and when you want to select the best agent for testing (this is how you clear a level)--\n\n![lvlClear.gif](\/\/\/raw\/deb\/a\/z\/3592b.gif)\n\nThe inputs the agent has available are 5 sensors that calculates distances to objects in front of it (see the lines). The agent's output is to rotate and\/or move forward. The performance (or fitness) of the best performing agent can be seen in the bottom right corner and is modulated by which factors rewards the agents in their actions.\n\nFor example, rewarding agents for the distance to the target means that the shorter the agents' distance is to the satellite at the end of the learning loop, the better its performance is. So be sure to decide how to reward the AIs or they might not even understand the task! :scream: \n\nEvery time 15 in-game seconds (can be scaled by you) go by, the agents are reset and the bottom half performing agents get updated neural networks through a mutation algorithm.\n\n![training.gif](\/\/\/raw\/deb\/a\/z\/35948.gif)\n\nWarning: There isn't any definite ending but just imagine this very elaborate ending where you realise that _you_ are an agent just like them - wooow :upside_down:\n\nI hope you will enjoy the game and I look forward to your feedback! :heart: \n\n```\nChangelog:\n5th October, 8:12: Known bug of disappearing agents after first iteration. Reason unknown. Added Windows build. Not working there either. Bug fix waiting.\n\n```\n\nPlay the game (HTML5) and see the source code here:","meta":{"author":[44013],"link-01-tag":[42336],"link-02-tag":[42332],"cover":"\/\/\/content\/deb\/a\/z\/3508a.png","link-01":"https:\/\/esbenkc.itch.io\/loop-to-learn","link-02":"https:\/\/github.com\/esbenkc\/LDJAM47"},"path":"\/events\/ludum-dare\/47\/the-learning-loop","parents":[1,5,9,212256],"love":0,"notes":5,"notes-timestamp":"2020-10-11T08:44:47Z","grade":{"grade-01":5,"grade-02":5,"grade-03":5,"grade-04":5,"grade-05":5,"grade-06":5,"grade-07":5,"grade-08":5},"magic":{"cool":0,"feedback":0,"given":0,"grade":5,"grade-01-average":3,"grade-02-average":3,"grade-03-average":4.5,"grade-04-average":4.333,"grade-05-average":4.667,"grade-06-average":4.333,"grade-07-average":2.167,"grade-08-average":4.333,"smart":-55.278640450004}},"text":"![title.gif](\/\/\/raw\/deb\/a\/z\/358f4.gif)\n\n## A Game of Reinforcement Learning\n\n[Play it here! :heart:](https:\/\/esbenkc.itch.io\/loop-to-learn)\n\nYou are an engineer tasked with optimizing artificial intelligent agents in the completion of spatial navigation tasks to satellites. You have control over the time scale, which aspects of the environment triggers rewards in the agents, and when you want to select the best agent for testing (this is how you clear a level)--\n\n![lvlClear.gif](\/\/\/raw\/deb\/a\/z\/3592b.gif)\n\nThe inputs the agent has available are 5 sensors that calculates distances to objects in front of it (see the lines). The agent's output is to rotate and\/or move forward. The performance (or fitness) of the best performing agent can be seen in the bottom right corner and is modulated by which factors rewards the agents in their actions.\n\nFor example, rewarding agents for the distance to the target means that the shorter the agents' distance is to the satellite at the end of the learning loop, the better its performance is. So be sure to decide how to reward the AIs or they might not even understand the task! :scream: \n\nEvery time 15 in-game seconds (can be scaled by you) go by, the agents are reset and the bottom half performing agents get updated neural networks through a mutation algorithm.\n\n![training.gif](\/\/\/raw\/deb\/a\/z\/35948.gif)\n\nWarning: There isn't any definite ending but just imagine this very elaborate ending where you realise that _you_ are an agent just like them - wooow :upside_down:\n\nI hope you will enjoy the game and I look forward to your feedback! :heart: \n\n```\nChangelog:\n5th October, 8:12: Known bug of disappearing agents after first iteration. Reason unknown. Added Windows build. Not working there either. Bug fix waiting.\n\n```\n\nPlay the game (HTML5) and see the source code here:","title":"The Learning Loop"}