Resources¶
Goal¶
provide a simple explination for the ML technical words thats are generally used in this site.
Labelled/Tagged Data¶
It means, we know the input type and outcome type,as they are tagged or data tags(labels) are available or, we know the data(ax+b) and the outcome(y)
used in supervised learning or generally called supervised learning.
eg: In image classification for cats vs dogs, this is both input(image) and output “cat” or “dog” is labelled.
Unlabelled data¶
No information/tags is available in/with the data or when the tagging(labeling) is expensive(time consuming).
used in clustoring and dimentionality reductions.
eg: finding hidden patterns in data.
Supervised Learning¶
Supervised learning is a model learning(like patterns, relationships etc) from a labeled data.
Since,labeled dataset contains a lot of examples of Features and Target,Models uses algorithms that learn the relationship of Features and model the Target from the dataset.
There are two types of supervised learning:
Classification
to classify the market to be bearish or bullish
Regression
to predict the share value of a apple stock
Unsupervised Learning¶
In unsupervised learning,since there is no labelled data, model learn the hidden pattern and relationship and starts grouping or clustoring the data.
Eg: unsupervised learning model can access can a customers who shop online tend to purchase multiple items from the same category.
But,this derived relation needs to be validated by the data scientist to cleary find the association between the X to Y.
Semi-supervised Learning¶
Combines a small amount of labeled data with a large amount of unlabeled data to improve learning accuracy.
Hyperparameters/HyperParams¶
Hyperparameters are external configuration settings, that are set by data scientists before training a machine learning model, rather than being learned by the model itself.
They control the behavior of the training process, defining how the model learns and influencing its complexity and speed.
eg: learning rate, batch size, or model architecture (layers/nodes).