Automatically build ML models to classify any data that can be represented in matrix form
AutoML Matrix is an algorithm which can automatically build state-of-the-art classification models using data presented in a matrix form (or tabular form). AutoML-Matrix is based on proprietary advanced Machine Learning algorithm. It’s available on Amazon Sagemaker Platform with features to train and deploy models at scale.
Data Preprocessing and Feature Engineering
Algorithm automatically transforms user data into a computationally desired form.
It identifies data points (rows in the data matrix) which are anomalous and discards them during model training.
Algorithm identifies features which are redundant or noisy and discards such features during model training.
It adds new features that are informative and useful to identify hidden patterns in the data.
Hyper-parameter Optimization and Training
User does not have to provide any Hyper-parameter values.
Algorithm has in-built fully automated hyper-parameter optimization.
AutoML Matrix uses fast and accurate proprietary learning algorithm.
Training stops automatically when it detects that the performance of trained model cannot be improved further.
Alternately, user can also provide a maximum training time, after which the training stops.
Machine Learning and Programming Resource Requirement
Using AutoML Matrix does not require the user to be a Machine Learning or Deep Learning expert.
The web console makes it very easy for domain experts with no programming experience to train and evaluate ML models.