Explainability and Interpretability
by Ashok Balasubramanian Updated: Dec 21, 2021
Will you trust a potential global war threat decision with AI? Reuters reported that Deputy Secretary of Defense Kathleen Hicks was briefed on a new software created by US military commanders in the Pacific that can predict Chinese reaction to US actions in the region. The tool looks at data since early 2020. It predicts response across various activities such as congressional visits to Taiwan, arms sales to allies in the region, or when several US ships sail through the Taiwan Strait. It is heartening to see AI mature into strategic roles, especially in the backdrop of Zillow iBuying algorithms causing a loss of more than $300m a few weeks ago and costing over 2000 jobs and an unsold inventory of 7000 homes! Well, the answer lies in strategic oversight. Algorithmic decisions reflect data quality, rigorous training, and introduced biases, among other factors. Both these situations reflect on the maturity of AI as a technology and the need for better design and review. With AI becoming almost a black box to most engineers given the complexity of the high number of parameters and nodes, Explainable AI brings a set of tools and frameworks to help understand predictions made by machine learning models. Explainability shows how significant each of the parameters and nodes contribute to the final decision. This helps debug and improve model performance and understand the model's behavior. Interpretability communicates the extent to which a cause and effect can be observed within a system. i.e. the extent to which you can predict what will happen, given a change in input or algorithmic parameters. Together, they help understand how the model arrived at a decision and how each step contributed to that. Try over 100s of Explainability and Interpretability solutions on kandi to make your next big decision.
A game theoretic approach to explain the output of any machine learning model.
Jupyter Notebook 18846 Version:v0.41.0 License: Permissive (MIT)
Lime: Explaining the predictions of any machine learning classifier
Fit interpretable models. Explain blackbox machine learning.
C++ 5378 Version:v0.3.2 License: Permissive (MIT)
A collection of infrastructure and tools for research in neural network interpretability.
Jupyter Notebook 4512 Version:v0.3.10 License: Permissive (Apache-2.0)
🔅 Shapash makes Machine Learning models transparent and understandable by everyone
Jupyter Notebook 2141 Version:v2.3.0 License: Permissive (Apache-2.0)
Quickly build Explainable AI dashboards that show the inner workings of so-called "blackbox" machine learning models.
Python 1578 Version:v0.4.2 License: Permissive (MIT)
DrWhy is the collection of tools for eXplainable AI (XAI). It's based on shared principles and simple grammar for exploration, explanation and visualisation of predictive models.
R 590 Version:Current License: No License