英文论文写作句子

1. However, deep learning models are
   vulnerable to adversarial attacks, in which deliberately designed small perturbations are added to the benign input samples to fool the deep learning model and degrade its performance
2. It improves the query efficiency by directly searching for informative examples in the input space for substitute model training, instead of taking a fixed step along the gradient direction.
3. Its main idea is to gradually reduce the magnitude of
   the adversarial perturbation while ensuring its effectiveness（其主要思想是逐步降低
   对抗性扰动，同时确保其有效性）
4. Albeit the outstanding attack performance of these methods,（尽管。。）
5. For the sake of brevity（为了简单起见）
6. Note that the real data is sampled from the true class, so as to encourage that the generated instances are close to data from the original class.
7. We defined and adapted two attacks, originally proposed for image recognition, for the TSC task
8. With deep neural networks becoming frequently adopted by time series data mining practitioners in real-life critical decision making systems we shed the light on some crucial use cases where adversarial attacks could have serious and dangerous consequences.（我们揭示了一些被对抗攻击可以造成严重后果的使用场景）
9. Adversarial examples are known to be transferable
   across different neural network architectures which enables the synthetic time series to fool other deep learning models: a technique known as black-box attack
10. One exception is the DiatomSizeReduction dataset
    which is the smallest one in the archive with an already low original accuracy equal to 30% due to overfitting
11. . Our preliminary experiments reveal that data augmentation can drastically increase deep CNN’s accuracy on some datasets and significantly improve the deep model’s accuracy when the method is used in an ensemble approach.
12. Time series classification models have been garnering significant importance in the research community.
13. Several black-box attacks that require no internal knowledge about the target systems such as gradients, have also been proposed

$$\begin{array}{cl}\underset{e(\mathbf{x}{)}^{\ast }}{\mathrm{maximize}}& f_{adv}(\mathbf{x}+e(\mathbf{x}))\\ \text{ subject to }& \Vert e(\mathbf{x}){\Vert }_0\le d\\ & \Vert e(\mathbf{x}){\Vert }_{\infty }\le ϵ\end{array}$$