Systems and methods for training machine learning models using active learning
Systems and methods for improvements in AI model learning and updating are provided. The model updating may reuse existing business conversations as the training data set. Features within the dataset may be defined and extracted. Models may be selected and parameters for the models defined. Within a distributed computing setting the parameters may be optimized, and the models deployed. The training data may be augmented over time to improve the models. Deep learning models may be employed to improve system accuracy, as can active learning techniques. The models developed and updated may be employed by a response system generally, or may function to enable specific types of AI systems. One such a system may be an AI assistant that is designed to take use cases and objectives, and execute tasks until the objectives are met. Another system capable of leveraging the models includes an automated question answering system utilizing approved answers. Yet another system for utilizing these various classification models is an intent based classification system for action determination. Lastly, it should be noted that any of the above systems may be further enhanced by enabling multiple language analysis.
1. A computer implemented method for training a model using active learning comprising:
collecting a corpus of conversations from a database;
applying one or more transformations to the corpus including entity replacement to create context, utterance and label sets, wherein the context, utterance and label sets are a first training set;
training a neural network using the first training set;
uploading a preset number of sentences;
suggesting high priority annotations in the uploaded sentences;
generating multiple annotations per action, intent, or entity found in the uploaded sentences;
select from the multiple annotations a subset of reliable annotations, where the subset is selected based upon low inter-annotator agreement;
creating a second training set comprising feature vectors defined using the subset of reliable annotations;
training a machine learning model using the second training set; and
generating an ensemble model by combining the trained neural network and the trained machine learning model using a loosely coupled metadata classifier, and wherein the ensemble model provides superior classification results compared to traditional machine learning techniques.
2. The method of claim 1 , wherein the training the machine learning model employs Amazon Web Services Spark.
3. The method of claim 1 , further comprising calculating F-score for the machine learning model.
4. The method of claim 3 , further comprising comparing the F-score to a threshold.
5. The method of claim 4 , wherein the threshold is 95%.
6. The method of claim 4 , further comprising outputting the machine learning model for production when the F-score is above the threshold.
7. The method of claim 4 , further comprising uploading a subsequent preset number of sentences and repeating annotation on said subsequent preset number of sentences to improve the machine learning F-score.
8. A computer implemented method for training a model using active learning comprising:
collecting a corpus of conversations from a database;
applying one or more transformations to the corpus including entity replacement to create context, utterance and label sets, wherein the context, utterance and label sets are a first training set;
training a neural network using the first training set;
uploading a preset number of sentences;
suggesting high priority annotations in the uploaded sentences;
generating multiple annotations per action, intent, or entity found in the uploaded sentences;
select from the multiple annotations a subset of reliable annotations as a second training set, where the subset is selected based upon low inter-annotator agreement;
training a machine learning model using the second training set, wherein the machine learning model includes an algorithm and a set of parameters; and
optimizing the set of parameters in a distributed computing system; and
generating an ensemble model by combining the trained neural network and the trained machine learning model using a loosely coupled metadata classifier, and wherein the ensemble model provides superior classification results compared to traditional machine learning techniques.
9. The method of claim 8 , wherein the optimization uses cross-validated grid-search.
10. The method of claim 8 , wherein the optimization uses randomized search cross validation.
11. The method of claim 8 , wherein the optimization uses sequential nested search cross validation.