1.記錄自己總結的規律,相信可以慢慢進入正軌。
2.爲了鍛鍊英文且不失去一些沒讀懂地方的原意,用英文記錄
目錄
Learning To Detect Unseen Object Classes by Between-Class Attribute Transfer
Zero-shot classification by transferring knowledge and preserving data structure
Zero-shot learning via discriminative representation extraction
-
Learning To Detect Unseen Object Classes by Between-Class Attribute Transfer
- Method:Based attribute
- Learning a attribute from a range of different classes
- Using:A attribute layer
- Problem:Need a lot of manually labeled training data
- Problem setting:
- Problem:learning with disjoint training and test classes
- Two method:
- DAP:uses an in between layer of attribute variables to decouple the images from the layer of labels
- Using the MAP(maximum a posterior) to predict unseen class
- 在實現時,爲什麼可以忽視p(z)
- p(z|x)=p(z|a)p(a|x)是什麼公式
- IAP:the attribute is from the connection layer between the classes are known or unknown
- DAP:uses an in between layer of attribute variables to decouple the images from the layer of labels
- Result
- DAP perform better than IAP
- Method:Based attribute
-
Zero-shot classification by transferring knowledge and preserving data structure
- Method:
- transferring knowledge and preserving data structure
- New Points:
- Transferred knowledge from source domain, the target classification model directly connects images and labels
- Use the attribute directly to learn semantic correction rather as a middle layer
- Preserving manifold data structure to handle domain shift problem
- ONE stage and solving the domain shift problem
- Manifold assumption
- If the samples are close in the low_dimension,they are also close in the origin space
- Like SAE
- Method:
-
Zero-shot learning via discriminative representation extraction
- Problem:
- Visual feature is not the best choice for zero-shot-learning because that it is for multi-shot task
- While the number of images more than classes,the mapping that maps all samples to one semantic space will not perform well
- Due to the hubness problem,inferring a test image in visual feature instead semantic space is not perform well;
- Point
- By discriminative zero shot visual representation
- Add a preprocessing stage before training stage
- Preprocessing stage:
- Reduce the dimension of the visual feature in order to get the template of each class
- Let the space be representation space
- Training stage
- Using the semantic embedding to learn a class template by a nonlinear regression.
- Testing stage
- First reduce the dimension
- Compare the similarity to decided which class belong to.
- Existing ZSL methods are based similarity and based projection
- Methods based on compatibility:
- Advantage: Take into account of the large margin
- Disadvantage: Don not make use of the neighborhood information
- Methods based on similarity:
- Advantage: Take use of the rich information of the seen class
- Do not meet the key factor to improve the accuracy
- Methods based on matrix:
- Can be seen as linear methods and also keep the information while transfer the information
- May ignore the nonlinear character reside on the dataset
- Contributions:
- Verify the supervised discriminant training on seen class can benefit the unseen class
- Aggregated perform better than Large Margin on ZSL
- Advantage:
- Directly inferring class in the visual representation space
- Problem: