updated gradient descent, underfitting, bais.md

terms/bias.md

@@ -3,8 +3,21 @@ references:
 - link_title: Bias of an estimator - Wikipedia
   link_url: https://en.wikipedia.org/wiki/Bias_of_an_estimator
 - link_title: Bias (statistics) - Wikipedia
-  link_url: https://en.wikipedia.org/wiki/Bias_(statistics
+  link_url: https://en.wikipedia.org/wiki/Bias_(statistics)
 related_terms:
 - variance
+- underfitting
 title: Bias
 ---
+
+Bias is the systematic error in prediction, caused by wrong assumptions in learning or by imbalances in training data. It reflects how far, on average, the model’s predictions deviate from the actual truth because of oversimplified or incorrect assumptions.
+
+Sources of Bias
+
+- Model assumptions (algorithmic bias): Using an overly simple model that cannot capture complex relationships.
+- Data bias: Training on data that is not representative of the real-world distribution.
+- Human or measurement bias: Errors in labeling, survey design, or feature selection.
+
+High bias lead to underfitting, and model is said to be too simple to capture the underlying pattern of the data.
+
+

terms/stochastic-gradient-descent-sgd.md

@@ -4,3 +4,13 @@ related_terms:
  - stochastic-optimization
  - gradient-descent
 ---
+Stocastic gradient algorithm picks a [[random single dataset example]][1] 
+or a subset of them at a time to calculate the values during [[backpropagation]][2] to 
+substract from models parameters, that is, weights and [[bias]][3] for updating them.
+
+How do it choose to pick one in random? [[Random.]][4]] 
+
+[1]: https://towardsdatascience.com/stochastic-gradient-descent-clearly-explained-53d239905d31
+[2]: /terms/Backpropagation
+[3]: /terms/Bias
+[4]: https://web.stanford.edu/class/ee270/scribes/lecture16.pdf

terms/underfitting.md

@@ -2,4 +2,19 @@
 title: Underfitting
 related_terms:
  - overfitting
+reference:
+ - link_title: Overfitting and Underfitting With Machine Learning Algorithms
+   link_url: https://machinelearningmastery.com/overfitting-and-underfitting-with-machine-learning-algorithms/
+ - link_title : Underfitting and Overfitting in Machine Learning
+   link_url :  https://www.baeldung.com/cs/ml-underfitting-overfitting
 ---
+
+Underfitting happens when the model is **too simple** to learn the important patterns in the data. 
+As a result, it performs poorly on both training and test data.
+
+### Signs of Underfitting
+- High training error
+- High test error
+- Predictions look overly simplistic compared to real data
+
+showing a very high bias is a sign of underfitting