So I've recently been interested in neural networks and i came up with a very basic implementation of them in factorio.
I'm only clamping the values because using sigmoid would require exponential calculations which can't be implemented directly
Schematic of the network:
(The numbers on the arrows represent the weight) In Factorio:
Code: Select all
H4sIAAAAAAAA/+2a4W6rIBTHX2Xx4w0uhapts8t9keXGWKUbiYUGcVljfPer7dpdQSnMpU1m+23Bc8Dz4xz+h5nxh5ynSf4Q44owSSUlBa4qlmwJ9lLOCpkw
6ad8u6YskVx4YMeL5inOcPWOfQgfQ7DHfvgY1iCjgqSHIQQaUyl4Hq/Ja/JGucDVhuaSiNZ5QV9YkuNK7nfNJG9UyDLJPXCc8zjor/OSeHXjpmQSB2gZrKIF
WkWAsoy8Y1jX7Rhjx/kap88e9P7iSpCsmeDwIvuYZu1CqEhLKts/YA3+G5p3h9of+HjvRFD5uiWSpsNvHli+9qevuBnM6Md6N1QUMnaKxJEFhoDviEgOE3q/
PMBLuStdXFmG7hDjZw/1jUXjYrdqYzefaOyCcbGLrhW7Zt2d0M2OvytFsJOdqJO4KyWA4EUQwrrxX3bNh2mEwwVipfr4QoG4ASj/yqSCYVJwZoFKC/MQqq43
ZyBoGomDDDjU0tODA2lBHsKxGM4cGIwiNZXMiQyoQgtU0DpzwmEeGUlpRsSFAwdepvHh6LtO6hOKZl27RLTrwt5vJxYnurt9fJCR8UbwbUxZY483SV4QS0wr
w0Zfqof5CYGOy8nNFzQBup6eKkhr7gDhBidRN8JqgHvSyVoyjCtuR11wh2UQXhBayAbr4hcaqiw0jKHhnLxYM4Ob1Ey/t2j+uUnR1KSftawDJnHYtxNmhp2k
9gqOuQqvnKtnlOhajaq13NPEo0M+oKlLCDV4YFBxmzWEmx/39uh+LEWGYmJqrdTbxL5CNbfNtWiUwrgLjIv9qXp/2UdLg2DVCCPDRG6N111DaE2wdYPbqYuq
UuyDbVCjSFWjX7nsnv3gy25klA2G1tbu6sEidONSQLkF6u58fJOdvzTUrsXl7TzYyuqPjlB1xy5nCnjGtJ/akz28tDrnKK2nwECL69xQslWl5aTG9GcN/+60
OsWnwGeMglKf7LBc9uJbDCOx/mrBQlzd5KMFaJCu6Bs+W7B4758pRkwtMqPvlPiyXBO/2CZ53o1beN4wlgaBq8HC1SByNVh+bnod+CkuZ9pjYKvNQw3KgjSu
c96kkhQlsWSpHbItL5oebJSEpJJsT6vqz4P6nIbAYNtTz0+G7QKeBJGlYA/xE2HZPxsMZHY8JQAA