A5040139490- Advanced Neural Network Applications
- Domain Adaptation and Few-Shot Learning
- Video Surveillance and Tracking Methods
- Advanced Image and Video Retrieval Techniques
Kyung Hee University
2021
In prevalent knowledge distillation, logits in most image recognition models are computed by global average pooling, then used to learn encode the high-level and task-relevant knowledge. this work, we solve limitation of logit transfer distillation context. We point out that it prevents informative spatial information, which provides localized as well rich relational information across contexts an input scene. To exploit propose a simple yet effective approach. add local pooling layer branch...
Data augmentation is an effective way to increase the diversity of existing training datasets that result in improved generalization ability convolutional neural networks (CNNs). The effect usually global for methods i.e., a single applied whole image, thus limiting local characteristics augmented images. Moreover, does not support most fundamental behavior CNNs they focus more on features (local texture, tiny noise etc.) than shapes. We refer this as bias property. In paper, we propose new...