Project (30%)
Descripton
For this project, in teams of two, you will be (1) identifying a machine learning problem to solve or a ML toopic to investigate, (2) reading two journal papers relevant to your project, (3) designing a solution and writing a program/simulation to solve it, and (4) summarizing your project (writing about your research problem, data, algorithms, solution and implementation, results, conclusions) in ~2-3 pages long conference style paper using LaTeX and the given AAAI style files. You may choose to work on a problem of your choice (subject to instructor’s approval), or you may select one from the list given by the instructor.
Timeline
All due dates are at class schedule.
(Step 1, 5%) Decide on project, to be approved by instructor – you must present your idea in class in 3min or less (“elevator speach”). Be clear, and rehears your brief presentation, you may use 1-2 slides. Note: 2/5 points will be class attendance when project is discussed. Present in class.
(Step 2, 5%) Show intermediate progress:
-
Code - about 30% of entire project
-
Paper – show pdf (generated from LaTeX) with at least three sections: introduction; background (including the discussion of the two journal papers); and problem description
-
Part of this grade is your attendance during all class sessions designated “work on project.”
Bring your work to class and be prepared to discuss it with the instructor.
(Step 3, 5%) Bring pdf paper which should be like final, i.e. project and results must be included. Examples of past papers are provided below. Paper must have:
-
Title, abstract, figures (or tables), data analysis, citations, conclusions, at least 2 references (the journal papers) which are explained/discussed as background or related research.
-
Formatting: use LaTeX and the following AAAI style files
- link to aaai.sty
- link to my example of maics11.tex file
- link to my example of maics11.bib file
- link to aaai.bst file
- link to aaai instructions from AAAI webpage
Below are listed eight criteria used to grade your paper. The instructor will give you feedback on this paper, which you must address/fix for your final submission. Bring pdf to class for an in-class peer review session.
(Step 4, 5%) Class presentation ~5 min and ~5 slides. You will be evaluated using the oral presentation rubric shown below. Bring ppt to class and present.
(Step 5, 10%) Final submission
-
Upload program and a picture that illustrate the output of your program (5%):
- Clear code (comments + modularization); correct implementation; thorough understanding of the underlying algorithms/methods
- Include correct citation if part of code is written by someone else; clearly point out your main contribution if building on existing software
- Picture ilustrates that the prigram ran on your computer
-
Upload paper (5%) - final paper must incorporate my suggestions.
Link for upload will be provided at class schedule; upload all files as a single zip file.
Criteria for paper
(C1) Topic selection (identify a manageable topic/project; NOT too general)
(C2) Using relevant existing knowledge, research, views (present information from relevant sources representing various points of view)
(C3) Design process (demonstrate skillful synthesis of materials read; NO misunderstanding of the theory)
(C4) Analysis (your ideas must be well organized and must reveal insights, patterns, differences, etc.)
(C5) Conclusions (present logical conclusions; NO unsupported conclusions)
(C6) Limitations and implications (discuss relevant and supported limitations)
(C7) Timely and correct submissions (if no staples ==> 1% off)
(C8) Grammar, content, correct formatting
Criteria for oral presentation
- Delivery Speak loud; make eye contact with audience.
- Content/organization Good slides (not too much text, has pictures/figures); clarity; 8 understand the topic well; answer questions well.
- Enthusiasm Show interest in the topic that is presented.
Possible topics
-
Investgate topics we have not covered like Gaussian Mixture Models, Expectation-maximization Alg., PCA, SVM, Data visualization, ANOVA, Adaboost; Check out this repository of videos in CS videolectures.net
-
Investigate other ML libraries: scikit-learn; Java DL4J; DeepLearning4J, deeplearning4j.org; Google’s TensorFlow for deep learning, https://www.tensorflow.org/; Google Cloud ML (costs $); AWS (Amazon Web Services)
-
Investigate latest related tech like generative AI and large language models; create a chatbot or look at other text analyses with ML techniques; Natural Language Prpecessing; sentiment analysis in Amazon reviews; e-mail spam filter
-
Use deep learbning to paint like Picasso, Van Gogh
-
See datasets at kaggle.com and Univ. of California-Irvine ML Repository for other ideas
-
Investigate Hadoop platform for large datasets and Mahout (explore Machine Learning algorithms on Hadoop)
-
Swarm intelligence; simulations; agents
Examples of past projects (pdf)
:
Example 1 - Leaf classification
Example 2 - Hadoop
Example 3 - A conference paper done with studnets
Teams
- … will be
- listed here
- etc.