Shypd.aiThe largest open index of AI toolsNeural-backed-decision-trees is an AI Agents & Automation tool that makes decision trees competitive with neural networks. It improves accuracy and generalization, offering interpretability for models on various datasets.
Trending
About
Neural-backed-decision-trees (NBDT) is an open-source project designed to enhance decision tree performance, making them competitive with neural networks. It achieves this by matching or outperforming modern neural networks on datasets like CIFAR10, CIFAR100, TinyImagenet200, and ImageNet, while also improving generalization to unseen classes by up to 16%. The tool offers a unique loss function that can boost original model accuracy by up to 2%. Users can convert their own neural networks into NBDTs, train them with a tree supervision loss, and perform inference using embedded decision rules. It provides quickstart options for running pretrained NBDTs, loading models in code, and generating various hierarchies (induced, WordNet, random) for customization and visualization.
Capabilities
Pricing & Plans
Open Source
Free
FAQs
Trending
