Project Topic Ideas


In case your team would like to look at some suggestions, here are a few possible project topics. Some of them involve external collaborators. These are usually Cal Poly alumni or people somehow connected to Cal Poly or me. For additional ideas, follow this link for some suggestions for possible topics from last year’s 481 class. You can also look at previous projects to get an idea for the topics other students worked on.

Games


Pac-Man's Ghost Behavior Algorithms

This is a very interesting article I came across on the ghost AI from Pac-Man, which is interesting in terms of both AI and game design. See http://gameinternals.com/post/2072558330/understanding-pac-man-ghost-behavior. [Suggested by Ryan Schmitt, 480-F10].

Genetic Algorithm for Cars

This is a very nice example of genetic algorithms for a complex optimization task. They use a 2D physics library and genetic algorithms to try and come up with little vehicles that can move the furthest. See http://megaswf.com/serve/102223/. [Suggested by Ryan Schmitt, 480-F10].

Neuroscience


Database project for Brain-Machine Interface data, brainliner.jp

This project is a collaboration with Makoto Takemiya, a Cal Poly alum who is now a PhD student in Japan.
The following is a brief description of the project he’s suggesting:

My project at work is a database project for Brain-Machine Interface data, brainliner.jp. This data consists of labels associated with brain activity patterns. The paradigm for using this data is to build some kind of statistical model, using SVM, for instance. Then given some novel activity pattern, the type of pattern can then be decoded using the statistical model. You can also combine multiple decoders to decode very complex stimuli, e.g., http://www.cns.atr.jp/dni/en/research-projects/reconstruction-visual-images/Our publicly available datasets are below:http://brainliner.jp/search/showall/1If you are interested in this, I can help you and your students find and understand a good dataset and perhaps recommend some data-mining tasks.We are also making a modular java application based on the netbeans platform to display and edit this data:https://github.com/ATR-DNI/BrainLinerIf you have any students interested in working with us on this open-source project, please let me know!

Natural Language Processing

Terminology Management System


Another Cal Poly alum, Sam Li, is now in China, working with a startup that employs a crowdsourcing approach to translation from Chinese to English (see http://www.zuodao.com; it helps to be able to read Chinese):

Something that we have been improving is our terminology management system. When we upload a document, the NLP system will break it down into segments and then we continue on by matching terms within the segments with an uploaded termbase file that contains translated terminology relevant to the document's topic.But here are some issues when it comes to matching the terms with the content of our documents: - Capitalization ("Hello", "hello", "HELLO" should both match as a term even if there's just "hello" in the terminology file). Exceptions to think about are proper nouns and such.- Abbreviations (ABB, A.B.B, A. B. B., A/B/B etc)- If the termbase contains two terms but with different translations, how should we handle this?- Need to be intelligent about handling plural/singular / as well as tense "Running, ran, etc". And of course a different set of issues will apply for other languages such as Chinese.


Information Extraction and Text Analysis

Together with a few others, I’ve been working on a system that extracts information from collections of text documents and helps human analysts to incorporate the extracted information into a knowledge structure that represents a model of the domain under consideration. This system uses natural language processing, in particular Named Entity Recognition (NER), for the analysis of the documents, an ontology for the domain model, and a service-based architecture that can be deployed in a cloud computing environment to enable flexible interaction between its components. Our first domain of interest is “terrorism financing”, which turned out to be a less than optimal choice: There is little information available about the way intelligence analysts work, and what kinds of tools they currently use. Also, the source documents available are very limited since people involved in terrorism financing typically don’t talk about this in blogs or other public venues, and documents available to intelligence services are typically classified. For more details, see two papers available from Cal Poly’s Digital Commons: We’re interested in expanding this system in a few ways, and there are some possibilities to explore such an information extraction system as a project in this class. Although we would prefer to coordinate the class project with our activities, this is not necessary, and you can build your own system. Here are some specific aspects of interest:
  • using the system or approach for a different domain; we are considering sustainability and cyber-security
  • adding components for the extraction of information from non-text documents (images, video, sound) to the system
  • analyzing sources with more real-time information such as twitter streams to better capture recent events
  • incorporating more sophisticated visualization options into the system

Semantic Web


Semantic Media Wiki



Semantic MediaWiki (SMW, see http://semantic-mediawiki.org/wiki/Semantic_MediaWiki) is a free, open-source extension to MediaWiki – the wiki software that powers Wikipedia – that lets you store and query data within the wiki's pages.Semantic MediaWiki is also a full-fledged framework, in conjunction with many spinoff extensions, that can turn a wiki into a powerful and flexible “collaborative database”. All data created within SMW can easily be published via the Semantic Web, allowing other systems to use this data seamlessly.

A possible project involving SMW is to use it for a repository of class projects and possibly other student work. Last year, two students used it for their Master’s theses: Christina Forney built one for her thesis on tracking sharks with an underwater autonomous vehicle (see http://wiki.cforney.org/), and Emily Schwarz for her plant identification system (see http://cslvm157.csc.calpoly.edu/plantidentificationwiki/index.php/Main_Page). This could also include a component to deal with lab and assignment submissions: As it is now, much of the work the students invest in the evaluation and analysis of AI systems (such as chatbots), is only seen by the instructor, although it may be of interest to the whole class, or even people beyond it. Ideally, I’d like to see a collaborative system where students can add to work done by others (from the same class, or earlier versions, or by people outside of Cal Poly), resulting in something like Wikipedia on a smaller scale and with a different focus.

Spatial Cognition

Over the last ten years or so, I’ve been working with the Spatial Cognition group at the University of Bremen in Germany. This includes about a dozen student internships where Cal Poly students spent the summer in Bremen to work with the group there. The most recent project was a Master’s thesis on "Conceptual Requirement Validation for Architecture Design Systems” by Greg Flanagan:

Computer-aided architectural design (CAAD) programs represent architectural design at a low level of spatial abstraction. While this representation model allows CAAD programs to capture the precise spatial characteristics of a design, it means that CAAD programs lack the underlying computational apparatus necessary to reason about design at a conceptual level.This thesis is a first step towards building a framework that bridges the gap between the conceptual aspects of a design and its low level CAAD-based spatial representation. Specifically, this thesis presents a new framework, referred to as the Conceptual Requirements Reasoner (CRR), which provides an architect with a framework to validate conceptual design requirements. The CRR will demonstrate how qualitative spatial representation and reasoning techniques can be used as the link between a design’s conceptual requirements and its underlying quantitative spatial representation.A museum case study is presented to demonstrate the application of the CRR in a real world design context. It introduces a set of museum design requirements identified in research and shows how these requirements can be validated using the CRR. The results of the case study shows that the CRR is an effective tool for conceptual requirements reasoning.


Especially if you are interested in a possible internship there, a project related to Greg’s thesis would fit very well.