相关实验视频
Updated: Jun 20, 2026

09:44
Methods to Test Visual Attention Online
Published on: February 19, 2015
11.8K
一种基于改进的自我注意模块的高分辨率方法来估计到达方向
Xiaoying Fu1,2,3, Dajun Sun1,2,3, Tingting Teng1,2,3
1National Key Laboratory of Underwater Acoustic Technology, Harbin Engineering University, Harbin 150001, China.
The Journal of the Acoustical Society of America
|October 22, 2024
概括
这项研究引入了改进的神经网络,用于在水下声学中高分辨率的到达方向 (DOA) 估计. 该方法提高了准确性和稳定性,即使在具有挑战性的低信号噪声比条件下.
科学领域:
- 水下声学 水下声学
- 信号处理 信号处理
- 机器学习是机器学习.
背景情况:
- 在水下声学中,高分辨率的到达方向 (DOA) 估计至关重要.
- 现有的子空间和稀疏表示方法在低SNR,有限的快照和高计算复杂性方面存在局限性.
- 神经网络方法看起来很有前途,但与大数据和传统结构扎.
研究的目的:
- 提出一种新的神经网络方法,以准确和强大的DOA估计.
- 解决水下环境中现有的DOA估计技术的局限性.
- 在具有挑战性的条件下提高性能,如低SNR和有限的数据.
主要方法:
- 开发了一个神经网络,包括一个改进的多头自我注意模块.
- 在注意模块中利用了大规模的卷积内核和残余结构.
- 引入了增强的输入功能,以处理不均的噪音和不平等的目标强度.
主要成果:
- 与稀疏表示方法相比,拟议的方法实现了更高的角度分辨率.
- 在低SNR和有限的快照下,在DOA估计中表现出极高的准确性和稳定性.
- 通过模拟和实验结果验证有效性.
结论:
- 改进的自我注意神经网络为高分辨率的DOA估计提供了优越的方法.
- 该方法有效地克服了传统技术在挑战水下声学场景方面的局限性.
- 提出的方法为DOA估计提供了强大而准确的解决方案.
相关概念视频
Application of Linearization and Approximation
A drone flying through complex terrain often relies on more than one sensing method to estimate small changes in altitude. Along with direct measurements, air pressure provides a useful indirect indicator of vertical movement. Atmospheric pressure decreases as altitude increases, and this relationship is commonly described using an exponential model. Although accurate, converting pressure measurements into altitude values requires calculations that are too complex to perform repeatedly during...
Maximizing the Directional Derivative
The directional derivative is a central concept in multivariable calculus that describes how a function changes at a given point when moving in a specified direction. This direction is represented by a unit vector, ensuring that only the orientation influences the rate of change. By varying the direction, different rates of change can be observed, demonstrating that the directional derivative depends strongly on the chosen direction.The directional derivative is computed using the gradient...

