The Python machine learning tools that exist, and problems with them
| Contact Me | This is perhaps the most robust and well-documented resource that I was able to find available. Strangely, though, I was unable to find it on python.org, which is the first place that many of the developers that I know go to look for Python libraries. In addition, while the documentation is thorough, it is difficult to read and wade through. One of the objectives of my project would be to create documentation understandable by even those with no experience with machine learning, so that they might be able to use the tools as a black box for their applications. PyAIML is the Python implementation of the popular Artificial Intelligence Markup Language (AIML), which has been used to create many chatbots and chatterbots. Unfortunately, this is the major difficulty that I have with this program - it is very specific to one application, and lacks the robustness of what I propose. However, this would certianly be included in whatever suite I develop, and is a good example of an intuitive set of functions that can be standardized across the library. This engine serves to aid in the simulation of cellular automata, most readily recognized in the classic example of Conway's Game of Life. Again, while this is relatively easy to understand and use, it is very specific to one application, which limits its general usefulness to the AI-developing community. In addition, it does not appear to be a very robust program, and there are a variety of noted optimization and standardization needs to improve it. MontyLingua parses English text and breaks it up into a semantic interpretation. While this has not yet been specifically applied to Aritificial Intelligence, I include it here because it could very easily be used in a tool to help programs understand user input. However, without that application included now, it is difficult for someone searching for machine learning resources to find. Similar to this and tagged for AI applications is the Natural Language Toolkit (NLTK), but is is reportedly not as accurate. |
Created by Andrew Bouchard, 2006