Final Project Feedback #3
Description
Total (22.5/24)
Title & Abstract (1.5/2)
• In the abstract, you should not include your results, unless you think these data are extremely important. Please be aware that this is extremely rare. You can check Arxiv for some articles on machine learning to see how they write their abstract. (-0.5)
Background (3/3)
Problem Statement (3/3)
Data (2.5/3)
• You need to describe all of your datasets instead of only presenting the link to your dataset. (-0.5)
Proposed Solution (3/3)
Metrics (2.5/3)
• Does not explain why metrics are appropriate. (-0.5)
Results (4/4)
Discussion (3/3)
• Interpreting the result (1)
• Limitations (0.5/0.5)
• Ethics & Privacy (0.5/0.5)
• Conclusion (1/1)
Other comments:
It’s actually not recommended to drop models before finding the optimized hyper-parameter for them. Searching for models greedily usually involves in training multiple models with different hyperparameters simultaneously. The speed of convergence is usually used to determine which model to drop. -Yicheng
Great job overall. Congrats on completing COGS 118A! -Jason