Question
If the size of the training set is too large relative to the test set, then a machine learning model will exhibit a high generalization error due to this phenomenon. For 10 points each:
[10h] Identify this phenomenon where a machine learning model corresponds too closely with the data it is trained on.
ANSWER: overfitting
[10e] Overfitting can occur in many machine learning models, one common example of which are these models named for having a structure inspired by the human brain.
ANSWER: neural networks [or neural nets]
[10m] The VC dimension measures how well an ML model can perform this task. Perceptrons perform the “binary” form of this task, which involves assigning data a label with two values.
ANSWER: classification [accept word forms like classifying; accept binary classification]
<Other Science>
Summary
| 2024 ACF Fall at Cornell | fall | Y | 9 | 11.11 | 89% | 11% | 11% |
| 2024 ACF Fall at Ohio State | fall | Y | 8 | 16.25 | 88% | 38% | 38% |
| 2024 ACF Fall at Georgia | fall | Y | 14 | 24.29 | 121% | 71% | 50% |
| 2024 ACF Fall at North Carolina | fall | Y | 9 | 12.22 | 67% | 11% | 44% |
| 2024 ACF Fall at Claremont Colleges | fall | Y | 5 | 22.00 | 100% | 20% | 100% |
| 2024 ACF Fall at Illinois | fall | Y | 9 | 21.11 | 100% | 44% | 67% |
Data
| Chicago A | Illinois D | 10 | 10 | 10 | 30 |
| Chicago B | Northwestern A | 0 | 10 | 0 | 10 |
| Illinois A | Iowa B | 10 | 10 | 10 | 30 |
| Purdue B | Indiana | 10 | 10 | 0 | 20 |
| Iowa A | Northwestern B | 10 | 10 | 0 | 20 |
| Notre Dame A | Illinois B | 10 | 10 | 10 | 30 |
| Illinois C | Purdue A | 10 | 10 | 10 | 30 |
| SIUE | WashU A | 0 | 10 | 0 | 10 |
| WashU B | Notre Dame C | 0 | 10 | 0 | 10 |