Mehmet Supervision Meeting - 24th August
===
###### tags: `Topic A` `Topic B` `Topic C`

:::info
- **Location:** Zoom
- **Date:** 24 August 2022 - 16:00
- **Attendees:**
- **Apologies:** None
:::

---

# Updates since the last meeting.

- Renewed myself
- Continued the paper but!
- Started to write a chapter
- Started to create a deep model dataset

# *"The Worry List"*

- Need a voltmeter-ish apparatuses
- https://www.msoon.com/high-voltage-power-monitor
- https://www.youtube.com/watch?v=UQKCCYEUMVs&ab_channel=AdafruitIndustries
- Not knowing what should the dataset contain?
- Not knowing if the article is worth publishing
- Publishing a paper(s)
- Any progress on the device list I gave you before the summer?
- There are some internal calls for funding which have come around in the last 7 days; if you re-forward the most recent versions of the files that will be very helpful.
- **MC** to send materials to **MF**
- Since IT did not give me admin privileges on my computer, I need a computer.
- Can I use my budget to buy a computer?
- If so, will I have full access to the computer?

# Short-term plans for next steps

- Finishing the writing a comprehensive chapter that contains all steps and experiments
- Completing the dataset. Need help for:
- What should the dataset contain?
- What should I pay attention to when performing experiments?
- Can you provide a measurement apparatuse?
- I couldn't find any software that shows the energy consumption per process. Is it possible to compare the percentage of the processor used by the process with the total energy consumption of the device?
- For example, assume that during the process the device consumes 300 mWh and uses 90% of the CPU. Can I say that the process consumes 270 mWh?
- My current approach:
- Train 1 epoch
- Use the same hyperparameters for all models as possible: learning rate, optimizer, etc.
- Since there are no datasets for the edge devices, I have decided to use CPU datasets. If I achieve to use benchmark tools, I will benchmark some single-board computers such as Jetson Nano, Raspberry Pi.
- Geekbench

# Deadlines on the horizon

- I intend to publish the article after completing the chapter.
- International Conference on Performance Engineering (ICPE)
- https://icpe2023.spec.org/
- **Abstract:** October 14, 2022
- **Paper:** October 21, 2022
- https://www.overleaf.com/project/620151a8c474ed10402d64fe

# Meeting notes

:::info
**Please note:** We will populate this section collaboratively during the meeting.
:::

-
-
-
-
-

# Actions
:::info
All of our actions should have the following things:
- An owner, the person leading on the activity.
- Sufficient detail so we can remember what we are supposed to do.
- A deadline for completion.
:::
:::info
These actions will appear in the next meeting notes either as an update since last time (if completed), on the worry list (if we have run into issues), or in the short-term plans (if we've had to push this item forward).
:::

- [ ] **MC** to send shopping list to **MF**
- [ ] **MF** to dig out some benchmarking methodology papers
- [ ]