Noise Data Augmentation for NLP Intent Classifiers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current chatbot systems face challenges in accurately classifying user intents due to inadequate training data, which leads to incorrect intent determination, especially when dealing with irrelevant context, and require costly manual annotation or outsourcing for labeled data.
Innovation Solution
The method involves augmenting training data with noise text from various sources, such as lists of words, text corpora, or dictionaries, incorporated into the original text at predefined ratios to generate augmented utterances, enhancing the resilience of the intent classifier to irrelevant context.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If training data is augmented with noise text, then the classifier's resilience to irrelevant context is improved, but the training data complexity increases
Solution Approach 1:
The patent applies noise text (irrelevant context) to training data, converting what would normally be considered harmful interference into a beneficial training mechanism. By intentionally adding noise text to training utterances, the classifier learns to distinguish relevant from irrelevant context, improving its resilience to noisy input during deployment.
Solution Approach 2:
The patent modifies the training data by changing its composition parameters - specifically by adding noise text at controlled ratios (e.g., 10-50% of total text). This parameter change in the training data structure enables the classifier to learn robust feature extraction that is invariant to irrelevant context.
2Measurement precision
If manual annotation is used to obtain labeled data, then the training data quality is improved, but the cost and time consumption increase
Solution Approach 1:
The patent enables the training system to self-generate augmented training data by automatically adding noise text to existing training utterances. This self-service approach eliminates the need for manual annotation of additional training data, as the system autonomously creates diverse training examples with controlled noise injection.
Solution Approach 2:
The patent creates augmented training examples by copying existing training utterances and adding noise text variations. Instead of manually creating new labeled data from scratch, the system replicates and modifies existing high-quality annotated data, preserving the original labels while generating diverse training variations.
3Measurement precision
If more training data is collected to improve classification accuracy, then the model performance is improved, but the data processing complexity increases
Solution Approach 1:
The patent performs preliminary data augmentation by adding noise text to training utterances before the actual model training process. This preliminary action prepares the training data in advance, allowing the model to learn from pre-processed augmented data without requiring complex real-time processing during training inference.
Data Source
AI summary
Techniques for noise data augmentation for training chatbot systems in natural language processing. In one particular aspect, a method is provided that includes receiving a training set of utterances for training an intent classifier to identify one or more intents for one or more utterances; augmenting the training set of utterances with noise text to generate an augmented training set of utterances; and training the intent classifier using the augmented training set of utterances. The augmenting includes: obtaining the noise text from a list of words, a text corpus, a publication, a dictionary, or any combination thereof irrelevant of original text within the utterances of the training set of utterances, and incorporating the noise text within the utterances relative to the original text in the utterances of the training set of utterances at a predefined augmentation ratio to generate augmented utterances.