1 / 4

2009 ML Project:

2009 ML Project:. Goal: Do some real machine learning… A project you are interested in works better Data is often the hard part (get it in plenty of time) Usually involves additional reading Using software packages OK, particularly comparing algorithms is fine

fjimmy
Download Presentation

2009 ML Project:

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. 2009 ML Project: Goal: Do some real machine learning… • A project you are interested in works better • Data is often the hard part (get it in plenty of time) • Usually involves additional reading • Using software packages OK, particularly comparing algorithms is fine • Insightful survey/review paper OK (may involve a longer talk) • Theory problem OK, but high-risk

  2. Requirements • Project proposals (due February 19) • 1 single-spaced page • Problem, data, and planned approach • Progress report (due March 24) • One to two pages • Brief, self-contained description of progress, • problems, and expected scope of final project • Short Presentation (April 7 or 9) • Final report (due April 11) • Body (~12 to 24 pages of text, 12 pt font) with references and citations. Can have appendices. • Should have good introduction/abstract (like a conference paper) and discuss related work

  3. Example Project Themes • Exploring techniques for file pre-fetching • Finding scene lighting from face images • Predicting senator’s votes based on campaign contributions • Customer and product clustering for recommender systems • Comparing Naïve Bayes and Neural nets for text classification • Adaptation in a virtual animal world using reinforcement learning • Road extraction from Aerial images • Predicting success of major league baseball teams • Exploring methods for handwritten digit recognition • Differences between ML and MAP estimators • Using Multi-dimensional Scaling for Dataset Visualization • A belief network tool predicting the presence and onset of a particular disease. • Predicting the future success/failure of graduate students

  4. Other Comments Individual Project • We have 36 students and we should have 36 different topics • You can not use things you did in the past PhD work as your project; however, it is perfectly alright to conduct a project that adds something to past work to conduct as part of your dissertation, MS thesis research. • Reference all sources you used in your final report including your own publications (not doing so is cheating!!!) • Some topics are too general to be completed in 6 weeks • Evaluating to which the techniques you developed work for the problem at hand is very critical for the success of the project. • Just using a piece of machine learning software might not be a good idea; comparing different approaches, on the other hand, is okay. • Where does the sudden popularity of hidden Markov models originate from? • Some project are too specific and it is not explained how the project fits into the general picture. • There has to be some learning in your project!! • Search, unless involves learning from past experience, is not a good topic.

More Related