相关实验视频
Updated: May 14, 2026

13:19
Deep Neural Networks for Image-Based Dietary Assessment
Published on: March 13, 2021
8.9K
神经网络的后门式水标的持久性:全面评估
概括
这项研究探讨了深度神经网络 (DNN) 水印对微调的稳定性. 一种新的方法通过重新引入训练数据,保护知识产权,恢复DNN的水标微调后.
科学领域:
- 人工智能的人工智能
- 机器学习 机器学习
- 计算机视觉 计算机视觉
背景情况:
- 深度神经网络 (DNN) 提供了先进的功能,但它们的资源密集型培训引发了知识产权 (IP) 的担忧.
- DNN通常在网上公开访问,需要强大的保护机制,如水印.
- 基于后门的水印是保护DNN的关键技术,但其对微调的稳定性仍然不确定.
研究的目的:
- 评估在经过微调的DNN中基于后门的水印的持久性.
- 提出和开发一种新的数据驱动方法,用于在微调后恢复DNN水标,而不暴露触发器设置.
- 调查重新引入水印修复培训数据的有效性.
主要方法:
- 在微调场景下对最近基于后门的水标持久性的广泛评估.
- 开发一种新的数据驱动技术,用于在微调后恢复水印.
- 使用损失景观可视化来理解水印恢复机制.
主要成果:
- 经过微调后,可以通过重新引入训练数据来恢复水印,前提是模型参数没有大幅变化.
- 触发器精度可以恢复到100%,这取决于使用的触发器样本.
- 在微调过程中引入训练数据可以帮助缓解水印消失.
结论:
- 拟议的方法提供了一个可行的解决方案,用于在微调后恢复丢失的DNN水印.
- 这种方法提高了基于后门的水标方案的稳定性和实用性.
- 进一步研究优化微调期间的数据引入,可以提高水标的持久性.
相关概念视频
Long-term Potentiation
Long-term potentiation, or LTP, is one of the ways by which synaptic plasticity—changes in the strength of chemical synapses—can occur in the brain. LTP is the process of synaptic strengthening that occurs over time between pre- and postsynaptic neuronal connections. The synaptic strengthening of LTP works in opposition to the synaptic weakening of long-term depression (LTD) and together are the main mechanisms that underlie learning and memory.
Long-term Depression
Long-term depression, or LTD, is one of the ways by which synaptic plasticity—changes in the strength of chemical synapses—can occur in the brain. LTD is the process of synaptic weakening that occurs over time between pre and postsynaptic neuronal connections. The synaptic weakening of LTD works in opposition to synaptic strengthening by long-term potentiation (LTP) and together are the main mechanisms that underlie learning and memory.
Long-term Potentiation
Long-term potentiation, or LTP, is one of the ways by which synaptic plasticity—changes in the strength of chemical synapses—can occur in the brain. LTP is the process of synaptic strengthening that occurs over time between pre and postsynaptic neuronal connections. The synaptic strengthening of LTP works in opposition to the synaptic weakening of long-term depression (LTD) and together are the main mechanisms that underlie learning and memory.
Hebbian LTP
LTP can occur when presynaptic neurons...
Hebbian LTP
LTP can occur when presynaptic neurons...
Net Change Theorem
The Net Change Theorem is a fundamental principle in calculus that establishes a direct relationship between a function’s rate of change and its accumulated change over an interval. Mathematically, it states that the definite integral of a function's derivative over a given interval [a,b] yields the net change in the original function:This theorem has significant applications in various real-world scenarios, including physics, economics, and engineering. A particularly useful application is in...

