Well, still similar to the Regularization applied to linear, the regularized cost function for logistic regression is defined by: $$J(\mathbf{w}, b) = -\frac{1}{m} \sum_{i=1}^m \left[ y^{(i)} \log(f_{\mathbf{w}, b}(\mathbf{x}^{(i)})) + (1 - y^{(i)}) ...
Since we can't be sure which of the parameter to penalize, we penalize all by adding the Regularization term to the cost function. A regularized cost function is defined by: $$J(\mathbf{w}, b) = \frac{1}{2m} \sum_{i=1}^m \left( f_{\mathbf{w}, b}(\mat...
So, to simply put, Regularization is just simply making your parameters smaller and relatively less effective to avoid overfitting. oh, and the beautiful thing? — you get to keep all your features because you don't know which one might just make the ...
I bring thee good news. There are ways of mitigating overfitting! Collect more training data. (but, but, but sometimes, collecting more data can be challenge, especially when they are not available) Next thing we might want to consider is to reduce...
So I have not been feeling well, but thankfully I just recovered. I'm just going to give a summary of this. Underfitting: This occurs when the model doesn't fit the training set well. We can also say that there's a high bias thingy going on "Normal...
I found Logistic Regression really interesting and it was quick for me to understand--I think that is largely because Linear Regression has created a mental map for me. It's kinda like if you understand a monkey, you'd not have a so much trouble unde...
So far, I have come a long way from Andrew Ng's Course on Machine learning Specialization. Well, since I'm just starting obsidian, I'm going to do a quick note of what I have learnt so far and continue from there onwards. Here are the things I've lea...
I recently spoke with a friend about the rigors of embarking on learning things. Her response to me was: “why don’t you just blog about it?” And so, that was my eureka moment for me to start blogging about my progress — my progress in diving into the...
