Path9.Mod1.e - Selecting Clustering, Anomaly Detection and Image Classification Algorithms for Azure ML Flashcards

Augmented learning https://learn.microsoft.com/en-us/azure/machine-learning/algorithm-cheat-sheet?WT.mc_id=docs-article-lazzeri&view=azureml-api-1 https://learn.microsoft.com/en-us/azure/machine-learning/media/algorithm-cheat-sheet/machine-learning-algorithm-cheat-sheet.png?view=azureml-api-1#lightbox

1
Q

The only clustering algorithm that comes with Azure ML

When you want to Discover a Structure in your data that allows you to separate similar data into intuitive groups using an unsupervised approach

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

PAD

When you want to Find and predict unusual or rare occurrences in your data, where fast training time is desired and the data can be summarized into correlated feature values

Explain what this algorithm is based on…

A

PCA-Based Anomaly Detection

Principal Component Analysis: Frequently used in exploratory data analysis because it reveals the inner structure of the data and explains the variance in the data.

PCA works by looking for correlations among components and determining the combination of values that best captures differences in outcomes. These combined feature values are used to create a more compact feature space called the principal components.

Most versions plot in 2-d to visualize all data, then works to identify clusters.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

OCS

When you want to Find and predict unusual or rare occurrences in your data by partitioning data by a single “boundary”, and your data has less than 100 features

A

One Class SVM. This algorithm establishes a decision boundary that separates the data space into two regions: one for the “normal” data points and another for the “anomalies” or outliers.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

RN DN

When you want to Classify an Image using modern, deep learning Neural Networks

A

Either ResNet or DenseNet

How well did you know this?
1
Not at all
2
3
4
5
Perfectly