Chatterbox Labs’ software products are built to work immediately within a standard Enterprise software environment. There is no requirement for bespoke hardware or software, all products can be deployed on commodity server hardware and run on the standard JVM.
It takes minutes to deploy our software inside your firewall. Updates can be pulled via secure online repositories in line with your business objectives.
Patented Explainable AI
Chatterbox Labs’ Explainable AI was built to alleviate an Enterprise’s frustration with black box outcomes.
Our XAI can work with any AI engine out of the box.
We address both business and scientific audiences, with three approaches:
- NLP Ablation. Answers the why in a business context
- Distance Measures. Audit & Trace for business users
- Influence Functions. For experienced Data Scientists
Our approach is not to replace your existing AI investments. Instead, our XAI sits as a layer on top of current IP assets with local and global explanations.
Patented Synthetic Data Generator
The Synthetic Data Generator (SDG) was built to:
- Alleviate the burden of preparing machine learning ready data using Patented Reinforcement Learning methods
- Delineate IP for data reusability & ownership using Sequential Regression with Multiple Imputation methods
- Create vast datasets for use in regulated environments using the NLP Text Generator
The SDG addresses 1) poor data, 2) small data and 3) no data.
SDG enables real-world Enterprise AI projects where data is the constraint.
Patented Cognitive Engine (AutoML)
Chatterbox Labs’ Auto ML for classification, regression and time series forecasting was built for Subject Matter Experts, Data Architects, Data Analysts and Software Architects with no formal machine learning background.
In order to solve real world problems across varied data types and industries, a plethora of machine learning methods are automated within the engine: Neural Network, Support Vector Machine, Naïve Bayes, Random Forest, Latent Dirichlet Allocation, LASSO, Ridge, Elastic Net, Exponential Smoothing & ARIMA.
The compelling attribute of our Auto ML is the subject matter expert, architect or analyst does not need to interact with any underlying machine learning. Operational execution is simple to absorb; a 3-hour workshop will provide the necessary training users require to deliver real-world AI outcomes.
The Data Validator accelerates the identification, validation and preparation of AI use cases by:
- Seamlessly connecting to data from multiple sources
- Validating the most appropriate ML task for each use case without coding
- Empowering Subject Matter Experts to propose the ML model of choice to business sponsors without any formal ML experience
- Assessing the viability of AI use cases in minutes