02_Hung-yi Lee_Pokemon Classification & strategy
Pokemon vs. Digimon
Function with unknown Parameters
Loss of a func(given data)
- Given a dataset
- Loss of a threshold
given data set
Training Examples
- If we can collect all Pokemons and Digimons in the universe
, we can find the best threshold - We only collect exampels
from
- if we can collect all Pokemons and Digimons in the universe
, we can find the best threshold
$$
h^{all} = \arg \min\limits h L(h, \mathcal{D}{all}) \qquad \text{理想}
$$ - We only collect some examples
from
we hope
Note:
model-agnostic
- don’t have assumption about data distribution
- any loss function can be used
Hoeffding’s Inequality:
- The range of loss
is [0, 1] is the number of examples in
To make P smaller: Larger
What if the parameters are continuous?
- Everything that happens in a computer is discrete.
- VC-dimension(not this course)
Tradeoff of Model Complexity
Strategy
Framework of ML
Training data:
Testing data:
pipeline
General Guide
Split ur training data into training set and validation set for model selection
Optimization issue
- Gaining the insights from comparision
这是 Residuals Network 论文上的结果
并不是overfitting,这代表着 56-layer的Optimization 并没有做好。
56-layer 的network,一定可以做到 20-layer 的泛化能力
Start from shallower networks(or other models which are easier to optimize)
If deeper networks do not obtain smaller loss on training data, then there is optimization issue.
Solution: More powerful optimization technology.
Overfitting
- Small loss on training data, large loss on testing data
- Data augmentation
Augementation要有道理,一般不会将图像上下颠倒
- constrained model
Bias-Complexity Trade-off
Cross Validation
过多的利用 Public Testing Set 去 select model,会使得模型很容易在公开测试集上过拟合。 因此不太推荐。
N-flod Cross Validation
适用于小模型
mismatch
- Your training and testing data have different distributions.
- Most HWs do not have this problem, except HW11
- Post title: 02_Hung-yi Lee_Pokemon Classification & strategy
- Create time: 2022-03-25 10:10:22
- Post link: Machine-Learning/02-hung-yi-lee-pokemon-classification-strategy/
- Copyright notice: All articles in this blog are licensed under BY-NC-SA unless stating additionally.