NCA-GENL - Data Analysis and Visualization (14% of the exam) - Section 4.1
Enhance datasets through data augmentation.
Enhance training datasets using data augmentation techniques - such as image flipping, cropping, colour jitter, and text paraphrasing - to artificially increase diversity and reduce overfitting. Recognise when augmentation is appropriate, and understand that augmentations must preserve the semantic label of the original sample.
Data augmentation
Practice question for this objective
lock_openFree sampleData Analysis and Visualizationeasy
A team is fine-tuning a text classifier but has only 500 labelled training samples. Which data augmentation technique is most likely to preserve the original label while increasing the number of training examples?
ARandomly shuffling the order of sentences within each document
BTruncating each document to exactly half its original word count
CTranslating documents into a different language and discarding originals
DReplacing selected words with synonyms drawn from a lexical databasecheck_circle Correct
Identify synonym replacement as a label-preserving text augmentation strategy that increases training data diversity. Synonym replacement draws on a lexical resource to swap individual words with meaning-equivalent alternatives. Because the substitution is meaning-preserving, the ground-truth label remains valid, and the modified sentence exposes the model to varied surface forms of the same concept. This directly addresses data scarcity without requiring new human annotations.
Why A is wrong: Shuffling sentences is tempting because it creates new document orderings, but it disrupts coherence and often changes the meaning enough to corrupt the original label, making it unreliable for classification tasks.
Why B is wrong: Truncation is tempting as a quick resize step, but removing half the content frequently drops sentiment cues or key facts, altering what the label should be and reducing information available to the model.
Why C is wrong: Translation alone changes the language domain, and discarding originals removes the source-language distribution the classifier needs; this is not standard augmentation but rather domain shift.
Why D is correct: Synonym replacement substitutes words with semantically equivalent alternatives, keeping the overall meaning and therefore the class label intact while producing a distinct token sequence the model has not seen before.
Answers that look right on this material and are not. Each one is a distractor from a different question in the NCA-GENL bank for this domain.
Catastrophic forgetting caused by retraining on the augmented corpus
errorWhy it is wrong: Catastrophic forgetting describes a model losing previously learned knowledge when trained on new data; it is a genuine fine-tuning concern but is unrelated to the label correctness problem described in the scenario.
They reduce the number of trainable parameters in the convolutional layers, lowering the model's capacity to memorise the training set.
errorWhy it is wrong: Tempting because reducing capacity does combat overfitting, but augmentation alters the input data, not the network's parameter count, so this misattributes the mechanism.
It adds genuinely new information to the dataset, improving the model beyond what the original labelled examples could ever support
errorWhy it is wrong: Tempting because more samples feels like more information, but augmentation only re-expresses existing examples; it adds variation, not information beyond what the original labelled data already contains.