做超市海报的网站,作网站流程,网站开发需要看什么书,寿光营销型网站建设出处#xff1a;IJCAI 2024 未开源#xff0c;链接貌似是#xff1a;jackyue1994/Sub-Adjacent-Transformer (github.com)
贡献#xff1a;1. 提出#xff1a;基于 “次邻域” 及 “注意力贡献” 的注意力学习机制#xff0c;以增强异常、正常的区分#xff1b;2. 首次… 出处IJCAI 2024 未开源链接貌似是jackyue1994/Sub-Adjacent-Transformer (github.com)
贡献1. 提出基于 “次邻域” 及 “注意力贡献” 的注意力学习机制以增强异常、正常的区分2. 首次将 “线性注意力” 及 “可学习的映射函数” 引入TSAD。 1. 基本思想
Time points usually have stronger connections with their neighbors and fewer connections with distant points. This characteristic is more pronounced for anomalies [Xu et al., 2022]. →
If we rely solely on subadjacent neighborhoods to reconstruct time points, the reconstruction errors of anomalies will become more pronounced, thereby enhancing their distinguishability. 2. 具体方法
the sub-adjacent neighborhoods (“次邻域” ) 概念
直观理解图(b)内的 2 - 1 区域1、2 是预定义的区域边界满足2 ≥ 1 0.
具体概念The sub-adjacent neighborhoods indicate the areas not immediately adjacent to the target point. 1、2 是预定义的区域边界满足2 ≥ 1 0. win_size 是划分的时序窗口大小. 红色部分代表 the sub-adjacent neighborhoods.
the sub-adjacent attention (“注意力贡献”) 概念
1. * 注意力贡献 (attention contribution)在同一窗口内将注意力矩阵的 “列” 视为各点对其他点的贡献
2. 计算每列各点的 the sub-adjacent attention 直观理解图 2 内虚线区域之和.
具体概念The sub-adjacent attention contribution is defined as the sum of particular non-diagonal elements in the corresponding column of the attention matrix.
3. 计算整个窗口内每列各点的 the sub-adjacent attention之和 4. 实际的细节问题如果目标点靠近的窗口起点或终点次邻域范围可能部分超出窗口因此可用的点数量会变少造成贡献不均不平衡(蓝色阴影部分) 解决方法
循环移位函数circular shift function通过对时间点进行循环移位确保边界点能够公平地参与注意力贡献计算即每个点的次邻域内始终有相同数量的时间点 线性注意力 损失函数 和 异常分数
1. 损失函数在损失函数中引入“次相邻” 的注意贡献引导模型关注 “次相邻” 的邻域 2. 异常分数 3. Dynamic Gaussian Scoring 3. 实验结果