Midterm Report
Evaluation and Policy

Team-based projects are integrated as an essential part of the course. These projects help students to get hands-on experience in applying different techniques studied in this course in the development of intelligent algorithms/systems. Students select one of the options listed below for which they will undertake an independent investigation and apply the studied algorithms in the course to solve a selected problem. Course instructor and TAs will provide support to help the students with searching and using the literature, analyzing the challenging aspects of the problem and writing the final paper of the project. The programming parts of the project must be implemented in Matlab/Octave.

Projects must be done in group of 4 students. Teamwork helps to achieve more than what could ever be achieved on your own, improve problem solving, foster creativity and innovation and improve decision making. Teams of students must conceive, design and develop the project. Different issues should be considered to form a collaborative team such as responsibility for assignment, team composition (hard and soft skills, previous academic performance, etc.) and the schedule of the team members (how easy to establish regular face-to-face meetings). Once team is formed, students should be willing to subordinate their personal preferences to the decisions of the team, and be willing to compromise in order to achieve a group consensus. As team work should have team rewards, team members will receive a common grade. However a free-riding team member will be penalized if a common and repetitive negative feedback (peer evaluations) received from the other team members. This feedback will be investigated before deciding the penalty.

To register, please fill in the registration form before Friday January 16, 2015 and submit it to your course instructor's eamil address. The following project options are available:

Option A (Application):
  •  Students identify a problem in a pertinent area of industrial and commercial importance such as surveillance, search and rescue, environment monitoring, industrial process control/monitoring, space exploration, pipeline assembly and inspection, intelligent transportation systems, rehabilitation, health care and home intelligence to name but a few.
  •  They explore the applicability of the different algorithms studied in the course to solve this problem.
  •  Students choose a machine intelligence approach to handle this problem to be implemented and justify their choice.
  • They analyze experimentally the performance of the implemented algorithm.
Option B (Comparative Study):
  •  Students will pick a challenging machine intelligence problem that interests them such adaptive speed/steering control, pedestrian detection and tracking, speech recognition, natural language understanding, voice emotion recognition, face recognition, facial expression recognition, predictive maintenance/analytics, learning motion primitives, PID auto-tuning, spam filtering, etc.
  •  Survey several approaches to tackle this problem.
  •  They conduct a comparative study between at least two of the surveyed approaches to quantitatively and qualitatively evaluate these approaches in terms of a number of well-defined evaluation metrics and preferably using a benchmark dataset if available.
Option C (Algorithm design):
  •  Students identify a machine intelligence problem for which there are no satisfying approaches. These problems include, but are not limited to, behavior/plan recognition, threat evaluation, context-awareness systems, situation awareness, cooperative mapping/mointoring, learning in multiagent systems, group formation, communication relaying of UAVs/AGVs, dynamic task allocation, and self-X capabilities of fully autonomous systems (self-configuration, self-management, adaptability, self-diagnosis, self-protection, self-healing and self-repair self-optimization, self-synchronization and self-organization).
  •  Develop a new technique to tackle this problem.
  •  Analyze theoretically and/or empirically the performance of their technique.

Course Project Paper:

The result of the course project will be a scientific paper (minimum 5 pages) along with part of the source code developed to solve a given problem. IEEE Manuscript Template must be used. The paper must contain the following:

Summary: The Summary should be a brief version of the full paper. It should give the reader an accurate overview. Be brief, but be specific.

1. Introduction: summarize the importance of the problem you are trying to solve and the reason that motivated you to select this project. Explain what was the problem or challenge that you were given? state the purpose of the project and how did you solve it? Enumerate the objectives of the project and describe in brief the structure of the paper.

2. Literature Review: Conduct a critical survey on similar solutions and explain how your solution extends or differs from these solutions.

3. Problem Formulation and Modeling: Include the problem statement and describe its model.

4. Proposed Solution: describe your proposed approach to solve the selected problem (DON'T include source code in the paper). Pseudocode can be used.

5. Performance Evaluation: Establish a set of evaluation metrics and run some experiments with different settings and/or values of algorithm parameters to quantitatively and qualitatively assess the performance of the developed solution. Students must identify the pros and cons of the experimented techniques and assess the quality of work as well as its fit with project objectives.

6. Conclusions: summarize the conclusion and future improvement. Explain how did you solve the problem, what problems were met? what did the results show? And how to refine the proposed solution?You may organize ideas using lists or numbered points, if appropriate, but avoid making your paper into a check-list or a series of encrypted notes.

References: Every paper needs references; in fact, your failure to consult references for guidance may be considered negligence. On the other hand, when you include sentences, photos, drawings or figures from other sources in your paper, the complete reference must be cited. Failure to do so is plagiarism, an academic infraction with serious consequences.

Project Delivery:

In order to complete evaluating the project, each team has to upload to designated DROPBOX on LEARN all materials related to the project (final course project paper according to the course policy mentioned above + a well documented Matlab/Octave code and executable code with ReadMe/UserGuide that shows how to install and use the developed software).