MACHINE LEARNING TECHNIQUES - LASA
MACHINE LEARNING TECHNIQUES - LASA
MACHINE LEARNING TECHNIQUES - LASA
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
13<br />
1.3.2 Crossvalidation<br />
To ensure that the splitting of data into training and validation sets was not a lucky split leading to<br />
optimal but non representative performance, one must apply crossvalidation. “Crossvalidation<br />
refers to the practice of confirming an experimental finding by repeating the experiment using an<br />
independent assay technique” (Wikipedia). In ML, this consists of splitting the dataset several<br />
times at random into training and validation sets. The performance of the algorithms is then<br />
measured on the whole set of experiments, e.g. by taking the mean and standard deviation of the<br />
error when applying the algorithm on each set. The standard deviation is important as it gives a<br />
measure of potential discrepancies in the structure of your data set. The crossvalidation phase is<br />
used to fine tune parameters of the algorithm (e.g. thresholds, priors, learning rates, etc) to<br />
ensure optimal performance.<br />
1.3.3 Performance Measures in ML<br />
1.3.3.1 Ground Truth<br />
Comparing the performance of a novel method to existing ones or trivial baselines is crucial.<br />
Whenever possible, one will use the ground truth as a means to compare performance of the<br />
system during training.<br />
In classification, the ground truth can be obtained for instance from manually labeling each data<br />
with the appropriate class label. If the choice of class label may be subjective (e.g. labeling facial<br />
expressions of happiness, sadness, etc), one may ask several raters to label the picture and one<br />
may take the average score across raters for the “true” label.<br />
1.3.3.2 ROC<br />
Graphical representations such as the Receiver Operating Characteristic, so called ROC curve<br />
provide an interesting way to quickly compare algorithms independently of the choice of<br />
parameters.<br />
The ROC curve is used in binary classification algorithms. It balances the so-called true positives<br />
(TP), i.e. instances when the algorithm classified correctly a datapoint to (y=+1) and the false<br />
positives (FP), i.e. instances when the algorithm classified incorrectly datapoints to (y=+1).<br />
Conversely, false positives (FP) and false negatives (FN) correspond to the above two cases with<br />
classification (y=-1).<br />
The ROC curve plots the fraction of true positives and false positives over the total number of<br />
samples of class y=+1 in the dataset. Each point on the curve corresponds to a different value of<br />
the classifier’s parameter (usually a threshold; e.g. a threshold on Bayes’ classification). Typical<br />
ROC curve is drawn below.<br />
© A.G.Billard 2004 – Last Update March 2011