Exploring the language of data, one algorithm at a time. Machine learning enthusiast, AI researcher, and advocate for tech-driven solutions to real-world challenges. Passionate about using AI for public safety and ethical technology. Join me as I document my journey through data, models, and insights.
Nothing here yet.
Feb 18, 2025 · 1 min read · Well, still similar to the Regularization applied to linear, the regularized cost function for logistic regression is defined by: $$J(\mathbf{w}, b) = -\frac{1}{m} \sum_{i=1}^m \left[ y^{(i)} \log(f_{\mathbf{w}, b}(\mathbf{x}^{(i)})) + (1 - y^{(i)}) ...
Join discussion
Feb 18, 2025 · 1 min read · Since we can't be sure which of the parameter to penalize, we penalize all by adding the Regularization term to the cost function. A regularized cost function is defined by: $$J(\mathbf{w}, b) = \frac{1}{2m} \sum_{i=1}^m \left( f_{\mathbf{w}, b}(\mat...
Join discussion
Feb 17, 2025 · 1 min read · 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 ...
Join discussion
Feb 14, 2025 · 1 min read · 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...
Join discussion
Feb 13, 2025 · 1 min read · 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...
Join discussion