Rotation-Invariant Restricted Boltzmann Machine Using Shared Gradient Filters
Giuffrida, Tsaftaris · cs.CV · 2016-06-23 · 原文
Finding suitable features has been an essential problem in computer vision. We focus on Restricted Boltzmann Machines (RBMs), which, despite their versatility, cannot accommodate transformations that may occur in the scene. As a result, several approaches have been proposed that consider a set of transformations, which are used to either augment the training set or transform the actual learned filters. In this paper, we propose the Explicit Rotation-Invariant Restricted Boltzmann Machine, which exploits prior information coming from the dominant orientation of images. Our model extends the standard RBM, by adding a suitable number of weight matrices, associated with each dominant gradient. We show that our approach is able to learn rotation-invariant features, comparing it with the classic formulation of RBM on the MNIST benchmark dataset. Overall, requiring less hidden units, our method learns compact features, which are robust to rotations.
讲义
讲义·推断 依据「原文」自动生成的结构化摘要（推断），非原文表述；以原文为准。
1. 人话版
Finding suitable features has been an essential problem in computer vision.
We focus on Restricted Boltzmann Machines (RBMs), which, despite their versatility, cannot accommodate transformations that may occur in the scene.
2. 领域脉络
本文类目：cs.CV，属于其所在研究脉络的最新进展。
3. 机制拆解
As a result, several approaches have been proposed that consider a set of transformations, which are used to either augment the training set or transform the actual learned filters.
In this paper, we propose the Explicit Rotation-Invariant Restricted Boltzmann Machine, which exploits prior information coming from the dominant orientation of images.
Our model extends the standard RBM, by adding a suitable number of weight matrices, associated with each dominant gradient.
4. 证据与数字
摘要未给出量化结果——留意原文的实验与数据。
5. 反例与边界
摘要未声明局限与反例——这是需要警惕的信号，精读时先问边界。
6. 跨领域连接与意外收获
思考本文机制能否迁移到你正在跟进的问题。
7. 可复用方法
把本文机制与你手头项目对照，找一个两周内能验证的最小实验。
8. 术语表
精读时把不熟的术语记入此处，作为下次回忆的锚点。