Fall 2013: Final Project
Deadlines:
- November 19: Final Project Proposal due (optional)
- December 10, 12, 16, and 18: Final Project oral presentations
- December 20 at 11:59 PM: Final Project written report due
Each student will be expected to propose an original research question, design and implement an experiment that answers the proposed question, analyze and report on the results. Original work performed for this project should include a theoretical component (mathematical analysis of algorithms and/or results), and an experimental component (experiments performed in matlab, or some other programming language, using real or simulated signals).
Each student will give an oral presentation of results (10 minute presentation, plus 3 minutes of questions = 13 minutes total), and a written description of the experimental methods and results. The written description should use the style files from a professional conference or professional journal in your field (e.g., ICASSP or the Transactions on Signal Processing), and should conform to all publication requirements of the target publication. The only significant difference between this final project writeup and a technical publication is as follows: your final project writeup should include background and derivations sufficient to demonstrate that your project involves the application of concepts learned in ECE 544. Thus, for example, if your project builds on an important derivation that you learned in ECE 544, then your final project writeup may reproduce the derivation (with comments sufficient to demonstrate your understanding), even if such reproduction would be inappropriate in an IEEE or ACM publication.
You may work together with other students on your final project, but each student will give a separate written and oral presentation. If you work together with other students, you should be prepared to demonstrate (1) that you understand all of the concepts that you present, and (2) that your personal contribution to the theoretical and experimental work is significant.
The canned/pre-designed final project used in previous years will not be available this year; you must design your own final project. If you have questions about the suitability of your final project design, please ask.
PresentationsDate and Time | Speaker | Title | Room |
---|---|---|---|
12/10 11:45 AM | Biplab Deka | Towards an accelerator architecture for deep learning | 204 TB (class) |
12/12 11:30 AM | Sean Yen | Classifying "beach" or "peach" based on event-related optical data (EROS) | 204 TB |
12/12 11:45 AM | Johannes Traa | Learning non-linear functions from datasets that contain variables on a circle (or a torus) | 204 TB |
12/12 12:15 PM | Glenn Ko | Source separation using Markov random fields and neural networks | 204 TB |
12/16 11:00 AM | Zach Stephens | Second order descent methods for estimating branch lengths in phylogenetic trees | 5239 BI (Beckman Institute) |
12/16 11:15 AM | Yang Zhang | Glottal estimation using a nonlinear ARMA filter | 5239 BI |
12/16 11:45 AM | Mijail Gomez | Improving neural network learning by modifying the output target function | 5239 BI |
12/18 09:00 AM | Igor Fedorov | Data quantization optimized for machine learning applications | 5239 BI |
12/18 09:15 AM | Bihan Wen | A study of clustered sparsifying transform learning | 5239 BI |
12/18 09:30 AM | Xianliang Kong | Dimensionality reduction and feature extraction of gamma-ray spectra | 5239 BI |
12/18 10:00 AM | Man Ki Yoon | Malicious code execution detection for real-time embedded systems using system call frequency distribution | 5239 BI |
12/18 10:30 AM | Huiguang Yang | Combining region and key-point based features for unsupervised object recognition from multiple images | 5239 BI |
12/18 11:00 AM | Ihab Nahlus | Different hardware implementations of support vector machines: a comparative analysis | 5239 BI |
12/18 11:15 AM | Minje Kim | Structured nonnegative matrix factorization for learning manifold of audio sources | 5239 BI |
12/18 11:30 AM | Amit Das | Transfer learning for cross-lingual automatic speech recognition | 5239 BI |
12/18 11:45 AM | Eric Kim | Autoencoder as a associative memory for hyper-dimensional computing | 5239 BI |
12/18 01:30 PM | Long Le | Application of Q-learning for POMDP in detection of Golden-Cheeked Warbler | 204 TB |
12/18 01:45 PM | Ulysses Lee | On the resiliency of neural networks in the failure of nodes; an erasure model | 204 TB |
12/18 02:00 PM | Greg Meyer | Neural networks for 3D face analysis | 204 TB |
12/18 03:00 PM | Dan Soberal | Speech/music classification using neural networks | 204 TB |
12/18 03:15 PM | Sai Zhang | Multiple classifier systems for addressing hardware error | 204 TB |
12/18 03:30 PM | Di He | Machine learning applied to vowel detection on continuous speech | 204 TB |
12/18 04:00 PM | Berat Levent GEZER | Classification of arrhythmic ECG beats using supervised learning algorithms | 204 TB |
12/18 04:15 PM | Ashish Khetan | 204 TB |