使用深度可分离的二进制和三进制神经网络实现关键词发现系统
Seongwoo Bae1, Haechan Kim1, Seongjoo Lee2,3,4
1School of Electronics and Information Engineering, Korea Aerospace University, Goyang-si 10540, Republic of Korea.
Sensors (Basel, Switzerland)
|July 8, 2023
概括
这项研究引入了一种用于关键词发现 (KWS) 系统的新硬件加速器,在单一设备上实现唤醒词识别和命令分类. 高效的设计大大减少了面积,提高了嵌入式应用程序的性能.
科学领域:
- 嵌入式系统工程 嵌入式系统工程
- 人工智能的人工智能
- 硬件加速器 硬件加速器
背景情况:
- 关键词发现 (KWS) 系统对于人机交互至关重要,通常需要单独的唤醒词 (WUW) 识别和语音命令分类.
- 对于KWS的深度学习模型,由于计算复杂性和对特定应用程序优化的需求,对嵌入式系统构成挑战.
研究的目的:
- 提出一个统一的硬件加速器,深度可分离的二元化/三元化神经网络 (DS-BTNN),用于同时WUW识别和命令分类.
- 通过优化神经网络计算,在嵌入式KWS系统中实现显著的区域效率.
主要方法:
- 开发了一个深度可分离的二元化/三元化神经网络 (DS-BTNN) 硬件加速器.
- 利用冗余的比特运算符来有效计算二元神经网络 (BNN) 和三元神经网络 (TNN).
- 在Xilinx UltraScale+ ZCU104 FPGA上实现了KWS系统,处理实时音频数据.
主要成果:
- 与集成单独的BNN和TNN模块相比,DS-BTNN加速器实现了49.3%的面积减少,在40纳米CMOS过程中最终面积为0.558mm2.
- 该系统表现出高精度:97.1%的基于BNN的WUW识别和90.5%的基于TNN的指挥分类在170MHz.
- 加速器在一个设备上高效地处理WUW和命令分类任务.
结论:
- 拟议的DS-BTNN硬件加速器为集成的KWS系统提供了一个高度面积高效的解决方案.
- 这种统一的方法显著降低了硬件复杂性,并改善了嵌入式语音控制应用程序的资源利用率.
- 演示的性能验证了二元化和三元化神经网络在资源有限的KWS环境中的有效性.
相关概念视频
01:21Association Areas of the Cortex
5.5K
Association areas are regions of the cerebral cortex that do not have a specific sensory or motor function. Instead, they integrate and interpret information from various sources to enable higher cognitive processes such as memory, learning, and decision-making. Some key association areas include the following:
Prefrontal Association Area: This area is located in the frontal lobe and is involved in planning, decision-making, and moderating social behavior. It connects with primary motor areas,...
Prefrontal Association Area: This area is located in the frontal lobe and is involved in planning, decision-making, and moderating social behavior. It connects with primary motor areas,...
5.5K
01:25Neural Circuits
1.3K
Neural circuits and neuronal pools are two of the main structures found in the nervous system. Neural circuits are networks of neurons that work together to carry out a specific task or process. They consist of interconnected neurons and glial cells, which provide structural and metabolic support.
Neuronal pools are collections of nerve cells with similar functions and interact through chemical and electrical signals. These pools include both interneurons (the central neural circuit nodes that...
Neuronal pools are collections of nerve cells with similar functions and interact through chemical and electrical signals. These pools include both interneurons (the central neural circuit nodes that...
1.3K