What is Explainable AI?

The field aimed at making machine learning models understandable to humans

A model could reject your application for a mortgage or diagnose you with cancer. These decisions have consequences. Serious consequences. Even if they are correct, we would expect an explanation. To get these we look to the field of explainable AI. 

Read on Medium

 Latest YouTube

🚀 Course 🚀 Free: https://adataodyssey.com/permutation-channel-importance Paid: https://adataodyssey.com/courses/xai-for-cv/ We dive into Permutation Channel Importance (PCI) and show you how to apply it using Python. We'll work with the Landsat Irish Coastal Segmentation (LICS) Dataset, a resource for advancing deep learning methods in coastal water body segmentation. This dataset includes 100 multispectral test images, each with a binary segmentation mask that classifies pixels as either land or ocean. This is a great start if you are interested in applying Explainable AI methods to remote sensing machine learning models. 🚀 Useful playlists 🚀 XAI for CV: https://youtube.com/playlist?list=PLqDyyww9y-1QA4-o4tTAF_iD5cKCC1qEA&si=oJSL6o40xNpJqqkT XAI: https://www.youtube.com/playlist?list=PLqDyyww9y-1SwNZ-6CmvfXDAOdLS7yUQ4 SHAP: https://www.youtube.com/playlist?list=PLqDyyww9y-1SJgMw92x90qPYpHgahDLIK Algorithm fairness: https://www.youtube.com/playlist?list=PLqDyyww9y-1Q0zWbng6vUOG1p3oReE2xS 🚀 Get in touch 🚀 Medium: https://conorosullyds.medium.com/ Threads: https://www.threads.net/@conorosullyds Twitter: https://twitter.com/conorosullyDS Website: https://adataodyssey.com/ 🚀 Chapters 🚀 00:00 Introduction 01:43 Landsat Irish Coastal Segmentation (LICS) dataset 02:54 Exploring LICS 09:58 Shuffling a channel 12:53 Applying PCI
Applying Permutation Channel Importance (PCI) to a Remote Sensing Model | Python Tutorial
🚀 Course 🚀 Free: https://adataodyssey.com/permutation-channel-importance Paid: https://adataodyssey.com/courses/xai-for-cv/ In this video, we dive into Permutation Channel Importance (PCI) — a powerful explainability technique that helps identify which color channels (like RGB in regular images) contribute to a machine leanring model's predictions. PCI is a simple yet effective approach within Explainable AI (XAI) and serves as a foundation for understanding more advanced computer vision methods, such as occlusion and SHAP, which also use permutations. 🚀 Useful playlists 🚀 XAI: https://www.youtube.com/playlist?list=PLqDyyww9y-1SwNZ-6CmvfXDAOdLS7yUQ4 SHAP: https://www.youtube.com/playlist?list=PLqDyyww9y-1SJgMw92x90qPYpHgahDLIK Algorithm fairness: https://www.youtube.com/playlist?list=PLqDyyww9y-1Q0zWbng6vUOG1p3oReE2xS 🚀 Get in touch 🚀 Medium: https://conorosullyds.medium.com/ Threads: https://www.threads.net/@conorosullyds Twitter: https://twitter.com/conorosullyDS Website: https://adataodyssey.com/ 🚀 Chapters 🚀 00:00 Introduction 01:59 Permutation in comuter vision 03:46 The PCI algorithm 06:14 Applications of PCI
Explaining Computer Vision Models with PCI
In this video, we dive deep into the world of anomaly detection with a focus on the Isolation Forest algorithm. Isolation Forest is a powerful machine learning model for identifying outliers in high-dimensional data, but understanding why an anomaly is detected can be a challenge. That's where SHAP (SHapley Additive exPlanations) comes in. We'll explore how to use both KernelSHAP and TreeSHAP to interpret the contributions of individual features to anomaly scores. You'll learn how to visualize and break down these contributions, making it easier to understand and explain the decisions made by Isolation Forest. This is particularly valuable in real-world applications like fraud detection, where knowing the 'why' behind an anomaly is just as important as identifying it. 🚀 SHAP Course 🚀 SHAP course: https://adataodyssey.com/courses/shap-with-python/ The first 20 people to use the coupon code "IFSHAP24" will get 100% off! 🚀 Free XAI Courses 🚀 Signup here: https://mailchi.mp/40909011987b/signup XAI course: https://adataodyssey.com/courses/xai-with-python/ 🚀 Companion article with link to code (no-paywall link): 🚀 [https://medium.com/towards-data-science/explaining-anomalies-with-isolation-forest-and-shap-0d5d1224b918?sk=38c81f496f2cd5fd31591aecd0bd2adf 🚀 Learn about Isolation Forests 🚀 https://www.datacamp.com/tutorial/isolation-forest 🚀 Useful playlists 🚀 XAI: https://www.youtube.com/playlist?list=PLqDyyww9y-1SwNZ-6CmvfXDAOdLS7yUQ4 SHAP: https://www.youtube.com/playlist?list=PLqDyyww9y-1SJgMw92x90qPYpHgahDLIK Algorithm fairness: https://www.youtube.com/playlist?list=PLqDyyww9y-1Q0zWbng6vUOG1p3oReE2xS 🚀 Get in touch 🚀 Medium: https://conorosullyds.medium.com/ Threads: https://www.threads.net/@conorosullyds Twitter: https://twitter.com/conorosullyDS Website: https://adataodyssey.com/ 🚀 Sections 🚀 00:00 Introduction 01:35 What is Anomaly Detection? 02:28 What is Isolation Forest? 05:57 Interpreting SHAP Values for Isolation Forest 07:44 Model Training 15:28 KernelSHAP with Anomaly Score 21:17 TreeSHAP with Average Path Length
Explaining Anomalies with Isolation Forest and SHAP | Python Tutorial
See more Subscribe