initial commit for cocktail-mash AI

README.md

@@ -1 +1,3 @@
-# Readme
+# cocktail-mash
+
+With [cocktail-mash-neat](./neat/README.md) you can watch a AI learning how to play cocktail-mash with NEAT.]

cocktail-mash-ai/.gitignore

@@ -0,0 +1,11 @@
+neat-checkpoint-*
+.idea
+.vscode
+__pycache__
+Digraph.gv*
+winner.pkl
+speciation.svg
+avg_fitness.svg
+neat-examples
+history
+*.ods

cocktail-mash-ai/LICENSE

@@ -0,0 +1,22 @@
+Copyright (c) 2007-2011, cesar.gomes and mirrorballu2
+Copyright (c) 2015-2017, CodeReclaimers, LLC
+
+Redistribution and use in source and binary forms, with or without modification, are permitted provided that the
+following conditions are met:
+
+1. Redistributions of source code must retain the above copyright notice, this list of conditions and the following
+disclaimer.
+
+2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following
+disclaimer in the documentation and/or other materials provided with the distribution.
+
+3. Neither the name of the copyright holder nor the names of its contributors may be used to endorse or promote products
+derived from this software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES,
+INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
+CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

cocktail-mash-ai/README.md

@@ -0,0 +1,109 @@
+## Interactive machine learning with Cocktail-Mash
+
+This is a machine learning implementation for Cocktail-Mash. 
+Cocktail-Mash is a game developed by inovex and can be controlled by 
+a AI trained by this NEAT implementation. 
+
+The process of training the AI can therefore be watched with this implementation.
+
+Additionally to the model a visualized image of the Neural Network and statistic of the learning process will be saved. 
+
+The best Neural Network and statistics supporting the understanding of the learning process are visualized at 
+the end of the learning process. 
+
+
+## How to setup:
+
+### Install all dependencies
+
+For visualizing the neural network ```graphviz``` must be installed at your system:
+
+```
+sudo apt-get install graphviz
+```
+
+All dependencies for this project can be installed by:
+
+``` 
+python3 -m pip install -r requirements.txt
+```
+
+## How to run:
+
+### Run a model already trained:
+
+Models saved to the disk can be run directly by providing their path to the ```neat_run.py```. For exammple to run the sample models provided in the repository use:
+
+```
+python3.6 neatrun.py model_ai_finish
+```
+
+If no path is provided the newest model from the history folder will be loaded.
+
+### Train a new model:
+
+To train a new model just start the training process with the following command:
+
+```
+python3.6 neattrain.py
+```
+
+All progress will be saved to the ```history/[date_time] ``` folder including: config, graph, network and checkpoints
+
+## Individualize the learning process
+
+There are two config files in this project ``` neat_config.json ``` for general setting and ``` config``` to change
+the behaviour of the machine learning process. By changing their settings you can take impact in the traing setup and process.
+
+### neat_config.json
+
+The ``` neat_config.json ``` specifies the learning setup of the game. The options are described in the following:
+
+``` numberOfEvolutions  ```
+The numbers of evolutions the model will take by training. Default is 100 the 
+
+``` runNewBest ```
+If a genome with a better fitness then all genomes of all generations before emerges run it.
+
+``` runEachGeneration ```
+Each n-th generation will be run in the browser so you can watch the training process.
+
+
+
+``` numberOfRandomRecipes```
+Denotes with which kind of recips the Model will be trained with. There are three options:
+
+- numberOfRandomRecipes > 0: The specified number of random recipes will be used to train the model 
+
+- numberOfRandomRecipes == 0: All 18 possible recipes of the game will be used for training
+
+- numberOfRandomRecipes < 0: All possible recipes plus abs(numberOfRandomRecipes) random recipes will be used for training
+
+
+```fillSteps``` The amouth of units filled in the glass each step. Default is 10.
+
+```viewNetwork``` True if the Network-Image should be shown after training. They will be saved in any case.
+
+```viewSpeciesGraph``` True if the Species-Graph should be shown after training. They will be saved in any case.
+
+```viewFitnessGraph```True if the Fitness-Graph should be shown after training. They will be saved in any case.
+
+```logToFile``` True if all output should be writen to a logfile. This is usufull if you want to run the training on a server and inspect the output later.
+
+### config 
+
+Only one option has an impact to our implementation all others are just for modifying the training process.
+
+##### num_outputs 
+ ```num_outputs``` specifies how many output nodes the model will have. 
+If there are 3 outputs the level will be finished by the algorithem as soon as the glass filling exceed 90%. 
+This is staticly programmed and thus the network doesn't has any impact on this decision. 
+If there are 4 outputs the fourth output will be used to decide if a level is finished. Thus the network is taking this decision by it's own.
+
+- num_outputs == 3: The Algorithem will stop the Level as soon as the Glass is filled by more than 90%.
+
+- num_outputs == 4: The Model itself will decide if when to stop the filling process
+
+
+All other options are best described in the [NEAT-documentation ](https://neat-python.readthedocs.io/en/latest/config_file.html). 
+Feel free to play around with them to learn about their impact to the models quiality. 

cocktail-mash-ai/config

@@ -0,0 +1,81 @@
+
+[NEAT]
+fitness_criterion     = max
+fitness_threshold     = 10000
+no_fitness_termination = True
+pop_size              = 200
+reset_on_extinction   = False
+
+[DefaultGenome]
+# node activation options
+activation_default      = relu
+activation_mutate_rate  = 0.0
+activation_options      = relu
+
+# node aggregation options
+aggregation_default     = sum
+aggregation_mutate_rate = 0.0
+aggregation_options     = sum
+
+# node bias options
+bias_init_mean          = 0.0
+bias_init_stdev         = 1.0
+bias_max_value          = 30.0
+bias_min_value          = -30.0
+bias_mutate_power       = 0.5
+bias_mutate_rate        = 0.7
+bias_replace_rate       = 0.1
+
+# genome compatibility options
+compatibility_disjoint_coefficient = 1.0
+compatibility_weight_coefficient   = 0.5
+
+# connection add/remove rates
+conn_add_prob           = 0.0
+conn_delete_prob        = 0.0
+
+# connection enable options
+enabled_default         = True
+enabled_mutate_rate     = 0.01
+
+feed_forward            = True
+initial_connection      = full_nodirect
+
+# node add/remove rates
+node_add_prob           = 0.2
+node_delete_prob        = 0.2
+
+# network parameters
+num_hidden              = 6
+num_inputs              = 7
+num_outputs             = 4
+
+# node response options
+response_init_mean      = 1.0
+response_init_stdev     = 0.0
+response_max_value      = 30.0
+response_min_value      = -30.0
+response_mutate_power   = 0.0
+response_mutate_rate    = 0.0
+response_replace_rate   = 0.0
+
+# connection weight options
+weight_init_mean        = 0.0
+weight_init_stdev       = 1.0
+weight_max_value        = 30
+weight_min_value        = -30
+weight_mutate_power     = 0.5
+weight_mutate_rate      = 0.8
+weight_replace_rate     = 0.1
+
+[DefaultSpeciesSet]
+compatibility_threshold = 3.0
+
+[DefaultStagnation]
+species_fitness_func = max
+max_stagnation       = 20
+species_elitism      = 2
+
+[DefaultReproduction]
+elitism            = 2
+survival_threshold = 0.2

cocktail-mash-ai/model_ai_finish/config

@@ -0,0 +1,81 @@
+
+[NEAT]
+fitness_criterion     = max
+fitness_threshold     = 10000
+no_fitness_termination = True
+pop_size              = 200
+reset_on_extinction   = False
+
+[DefaultGenome]
+# node activation options
+activation_default      = relu
+activation_mutate_rate  = 0.0
+activation_options      = relu
+
+# node aggregation options
+aggregation_default     = sum
+aggregation_mutate_rate = 0.0
+aggregation_options     = sum
+
+# node bias options
+bias_init_mean          = 0.0
+bias_init_stdev         = 1.0
+bias_max_value          = 30.0
+bias_min_value          = -30.0
+bias_mutate_power       = 0.5
+bias_mutate_rate        = 0.7
+bias_replace_rate       = 0.1
+
+# genome compatibility options
+compatibility_disjoint_coefficient = 1.0
+compatibility_weight_coefficient   = 0.5
+
+# connection add/remove rates
+conn_add_prob           = 0.0
+conn_delete_prob        = 0.0
+
+# connection enable options
+enabled_default         = True
+enabled_mutate_rate     = 0.01
+
+feed_forward            = True
+initial_connection      = full_nodirect
+
+# node add/remove rates
+node_add_prob           = 0.2
+node_delete_prob        = 0.2
+
+# network parameters
+num_hidden              = 6
+num_inputs              = 7
+num_outputs             = 4
+
+# node response options
+response_init_mean      = 1.0
+response_init_stdev     = 0.0
+response_max_value      = 30.0
+response_min_value      = -30.0
+response_mutate_power   = 0.0
+response_mutate_rate    = 0.0
+response_replace_rate   = 0.0
+
+# connection weight options
+weight_init_mean        = 0.0
+weight_init_stdev       = 1.0
+weight_max_value        = 30
+weight_min_value        = -30
+weight_mutate_power     = 0.5
+weight_mutate_rate      = 0.8
+weight_replace_rate     = 0.1
+
+[DefaultSpeciesSet]
+compatibility_threshold = 3.0
+
+[DefaultStagnation]
+species_fitness_func = max
+max_stagnation       = 20
+species_elitism      = 2
+
+[DefaultReproduction]
+elitism            = 2
+survival_threshold = 0.2

cocktail-mash-ai/model_ai_finish/neat_config.json

@@ -0,0 +1,11 @@
+{
+        "numberOfEvolutions":100,
+        "runEachGeneration": 0,
+        "runNewBest": true,
+        "numberOfRandomRecipes": 0,
+        "fillSteps":5,
+        "viewNetwork": true,
+        "viewSpeciesGraph": false,
+        "viewFitnessGraph": false,
+        "logToFile": false
+}