Yi-Chen (Howard) Lo

### Relabeling Details #### Network architecture modification for generating label map  * Original: Feature map [H, W, d] => Global Pooling [1, 1, d] => Predicted label [1, 1,...

### Training with LabelPooling Once label map is pre-computed, we can train a new network by the following procedure: 1. Load image & label map (15x15xC) 2. Augmented image =...

### TL;DR Present a simple module capable of learning arithmetic functions such as add, sub, mult, div, etc. And can generalize well on unseen data or unseen inference scheme. ###...

### Summary - This paper present a new multitask question answering network (MQAN) that jointly learns all tasks in ten different NLP tasks. (Cast 10 tasks to question answering) -...

### Types of Question - Object recognition - What is in the image? - Object detection - Are there any cats in the image? - Attribute classification - What color...

### Abstract   - Learn an unconditional (i.e., generate from noise) generative model, SinGAN, that captures the internal statistics of a **single** training image. - SinGAN can generate new...

### TL;DR 1. This paper aims to translate the success of language model pre-training from texts to images by proposing the BERT-like self-supervised learning method called Selfie. 2. Selfie combines...

@gpuZheng Hi, sorry for the late response. It seems that the Mega automatically removed our dataset... The reason may be that our account is not premium. For quick solution, I...

@star-lzy The pretrained models are simply just imagenet pretrained backbone. You can also found in [FC4 codebase](https://github.com/yuanming-hu/fc4/tree/master/models/fc4).

@sid972 Hi, I think you can also save iteration/epoch step in checkpoint, then reload them if you wanna retrain the model.