Adder is meant to be a developer-friendly and easy-to-learn library that provides implementations of many AI algorithms (and a few other that ease working with graphs and trees in general).
Take a look at the wiki to get started.
Here's where the library currently stands:
- Utilities for defining various models of problems (from graph, from a set of functions, etc.)
- Algorithms for classical searching:
- BFS, DFS
- Depth-limited search, Iterative Deepening Search
- A*
- Algorithms for nonclassical searching:
- Hill climbing, Random-restart
- Simulated Annealing
- Genetics
- Classical Propositional Logic:
- Parsing sentences into formulas in Conjunctive Normal Form (thus having a human-readable representation of the knowledge base)
- Forward and backwards chaining on Definite Knowledge Bases
- Theorem proving in random Propositional Knowledge Bases with several optimizations
- Classical First-Order Logic:
- Backwards chaining for Definite Knowledge Bases
- General theorem proving for First-Order Knowledge Bases
- Fuzzy Logic:
- Inference
- Standard fuzzy sets - triangles, shoulders.
Currently, there are several demos you can play with. Run them with
python main.py <demo-name> [<demo-args>]where
<demo-name> = snake | picture | definitekb | resolutionkb
<demo-args> = <snake-field-size> <snake-obstacles-count> | one of the three - winnie toucan eagle | no args | no args
- Probabilistic Knowledge Bases
- Machine learning