Learning non-maximum suppression Supplementary material

Supplementary Material AMTnet: Action-Micro-Tube Regression by End-to-end Trainable Deep Architecture

In all our experiments, at training time we pick top 2000 RPN generated 3D proposals using NMS (non-maximum suppression). At test time we select top 1000 3D proposals. However, a lower number of proposals, e.g. top 300 proposals does not effect the detection performance, and increase the test time detection speed significantly. In Section 3.2, we show that extracting less number of 3D proposals...

Supplementary Material: Jointly Learning Non-negative Projection and Dictionary with Discriminative Graph Constraints for Classification

The JNPDL model is well motivated by the current drawbacks of dictionary learning approaches, while each constraints are also well designed (the novel discriminative graph constraints are proposed, and all constrains are designed to be easily optimized). Aiming to bridge the gap between features and dictionary, we do think the proposed idea of learning a projection for features jointly with the...

Supplementary Material: Unconscious numerical priming despite interocular suppression

Inverse Reinforcement Learning with Simultaneous Estimation of Rewards and Dynamics - Supplementary Material

This document contains supplementary material to the paper Inverse Reinforcement Learning with Simultaneous Estimation of Rewards and Dynamics with more detailed derivations, additional proofs to lemmata and theorems as well as larger illustrations and plots of the evaluation task. 1 Partial Derivative of the Policy

Supplementary Materials for Learning to Recognize Discontiguous Entities

This is the supplementary material for “Learning to Recognize Discontiguous Entities” [Muis and Lu, 2016]. This material gives more details in the experiment setup, the ambiguity of each model, and compare the models from theoretical point of view.