Sample Integrative Questions and Rules for Integrative Paper

Rules and Guidelines for Final Integrative Paper

• Each student will address a particular question. Questions below are representative; other questions are possible for discussion, but must be cleared beforehand.
• The paper needs to be between 9-11 pages (including references), two-column, single spaced, IEEE Transactions format. Figures must be included in the text and in the 9-11 pages.
• The paper is not a literature review, but a clear rhetorical argument towards a particular technical position.
• The paper must have least 6 significant diagrams / illustrations that summarize or argue for your point. These are diagrams that you generate. You can use additional diagrams from other sources, but you must reference them.
• A list of references, with indications in the text, is required.
• The paper is due on April 22, 2024, at 11:59pm. This date is the last day of regular class for our semester. Late papers will not be accepted.
• Submission will be a .pdf file, submitted to me by email for each assignment.

• Monday, April 1 (9:04am): Topic selected and where necessary, approved. One key figure drawn (10 percent).
• Monday, April 8 (9:04am): Paper Outline. Two additional key figures drawn (10 percent).
• Monday, April 15: Full paper drafts brought to class. One paper, one sent by email to me by 9:00am (20 percent).

Representative Papers for Discussion

Required reading

• Std Cell Intro

Additional Papers

Suggested Paper Topics

• Topic 1: NonVolitile Synaptic Devices Many believe that nano-devices (memristors, RRAM, etc) will enable building neural systems. Several have stated a silicon brain will never be built without these technologies. Others have stated that silicon not only is sufficient for building Si Cortex, but is the leading technology for building Si Cortex. Your assignment is to analytically work through this debate, and report your particular conclusions. This paper is not primarily a review paper, although you might need to give a short review of details required.
• Topic 2: Building a Synthetic Brain Several authors discuss the possibility of building a human brain in artificial technology. One sees a lot of hype in these areas, particularly in the popular press. The hype is large enough that Europe created a 1 Billion Euro, 10 year project. In the US, there is also large resources being considered. The opportunities are significant for building brain like structures, which you will want to summarize in your paper. The primary question to address is what technical directions should be the focus of these resources?
• Topic 3: Starting from what you see in class about silicon neurons in class, work through neuron models for physical computation. Do not give a review paper. There are a couple of papers that are review papers, and these papers did not a particular good job of a review, although it presented a huge amount of raw information). You need to give an analysis of forms of silicon neurons, the biological modeling and the resulting silicon implementation. You must deal with real types of inputs; only comparing with rate encoded signals is not acceptable.
• Topic 4: Dendritic computation is highly discussed in the neuromorphic literature. Give a sense why these issues are so debated, the issue of resulting computation required, particularly digitally, A short summary of potential computation would be appropriate. Key question is what computation has the highest likelihood of having impact.
• Topic 5: Devices for long-term structures. Biological systems operate over a wide range of timescales, from 10s of microseconds to seconds, to minutes, hours and longer. Biology uses longer-term timeconstants for regulation. The question is what devices and approaches are possible for these long range of timeconstants, and what are the power requirements. These timeconstants need to happen with very low power requirements.
• Topic 6: Question of neuromorphic computation as one form of physical computation. What computational properties are possible in physical computation, particularly as a question of continuous variables versus computing over discrete variables in typical digital computers.
• Topic 7: Question of resolution for synaptic weights. Question of noise and mismatch in neuro systems, bio and synthetic.
• Topic 8: Discuss computation in the human (or related mammal) auditory pathway and discuss the potential physical implementation of a useful computation (e.g. Speech Recognition). Describe the biological structures needed; just doing a deep learning network is not sufficient.
• Topic 9: Discuss computation in the human (or related mammal) visual pathway and discuss the potential physical implementation of a useful computation (e.g. Moving object recognition). Discuss the complexity issues for a physically implemented system. Describe the biological structures needed; just doing a deep learning network is not sufficient.

You are encouraged to use references discussed in class, as well as outside reference materials. Mostly you will need technical papers of various forms. Some references from on-line sources (e.g. Google search) are allowed where appropriate, but should not outnumber your technical sources.