TruEra Open Sources ‘TruLens’, A Cross-Framework Python Library For Deep Learning Explainability

6330
Source: https://github.com/truera/trulens

TruEra, the provider of a suite of AI quality solutions, release ‘TruLens‘ – an open-source explainability tool for machine learning (ML) models based on neural networks.

TruLens offers the ability to have constant explanations and insights into how deep neural network systems work by providing a uniform API for explaining TensorFlow, Pytorch or Keras-based model versions without having to switch between different libraries.

TruLens is a powerful new explainability solution for neural networks that offers deep insights and understanding into why the network arrives at certain conclusions. It provides an easy way to understand what concepts within images are being used by facial recognition models or medical diagnostic models, so users can better make sense of their decision-making.

TruLens can be used across a wide range of real-world use cases to reveal the inner workings of deep learning models. Use it for computer vision, natural language processing, forecasting and personalized recommendations.

The quality of AI in the lab and real-world is driven by explainability. TruLens provides a precise way to understand how these models are performing, allowing developers to refine their models better before they go live or fix any issues once used on the production line.

The library also provides support for a set of other popular explainability techniques created by the research community, including Saliency Maps, Integrated Gradients, and SmoothGrad.

TruLens is now available at : https://www.trulens.org/

Github: https://github.com/truera/trulens

Installation: https://pypi.org/project/trulens/

Paper: https://arxiv.org/abs/1802.03788

Source: https://www.globenewswire.com/news-release/2021/08/10/2277999/0/en/TruEra-Open-Sources-TruLens-Neural-Network-Explainability-for-ML-Models.html