zerohd4869

> Thank you for your work !! > > I want to train my dataset, what data format is included in the iemocap_features.pkl file of the dataset ? Is there...

> I found that the MELD dataset you use is 5-classified, and there is no comparison of how you process the two types of data in feature extraction, or how...

1. CAT的处理包含两部分：1）生成对抗扰动r。这部分代码可参考adversary_flag=True的代码块（at_method默认为FGM）；2）将扰动r加在建模上下文的hidden layer的权重上，之后再基于原优化目标进行梯度更新。这部分代码主要体现在emb_names变量的设置中，指定扰动r起作用的位置。如论文的图2(a)所示, 在输入信号u经过的多通道网络中，与u相关的所有带参数隐层均加放入了emb_names变量中。 2. 对于emb_names的选择：如果是上下文相关的输入，可使用LSTM/GRU等序列网络来建模上下文依赖，可直接参考我们的设置；如果是上下文无关的输入，如使用BERT建模单个句子，可直接将扰动放在embedding层或者最低的1-2层 hidden layer即可。 English version: 1. The processing of CAT consists of two parts: 1) Generating an adversarial perturbation r. This part of the code...

> Is pgd function not used? Is adversary_flag True when it is officially run? For the results in the paper, CAT is implemented based on FGM. The adversary_flag is set...

Thank you for your attention to our work. If you have any questions during the use, please feel free to communicate with us. We will do our best to provide...

> This work is exceptionally insightful. As a newcomer to this field, how should I go about applying your code to other datasets? Our work is a two-stage training process...

Thank you for your interest to our work. If you have any questions while using our services, please feel free to reach out to us. We will strive to offer...