Please enable JavaScript.
Coggle requires JavaScript to display documents.
Super vision Learning, Neural Networks, Methodologies, Clustering - Coggle…
Super vision Learning
Classification
-
-
-
-
Linear Classifiers
-
-
-
-
-
-
-
Multi class
1 vs All
-
-
-
Decision Trees
Contingency Tables
-
-
-
-
-
Information gain of Y given a split A={l,r}
$$IG(Y|A) = H(Y) - \sum P_aH(Y_a)$$
Pa proportion of the data in branch a \( \in \)
-
Over-fitting
-
-
$$E(M,D{train}) < E(M', D{all})$$
$$E(M, D{all}) > E(M', D{all})$$
-
-
-
-
-
Naive Bayes
-
-
-
-
-
Over fitting
-
-
-
-
-
Neural Networks
Gradient descent
\( \hat{\theta} = \arg\min_{\theta} \frac{1}{m} \sum_{i=1}^m LCE(f(x(i); \theta), y(i)) \)
where LCE is the loss function, f is the model, and y is the label.
-
steps
-
Step2 : Calculate the weight values w into in terms of the
derivative of the loss function for each separate wj
-
Step 4: Calculate the new parameters, as:
New parameter = old parameter - step size
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Linearly Non-Separable Problems
E.g., the famous XOR problem
-
-
-
-
