|$\propto$SVM for learning with label proportions
- Learning setting: Only the proportion of each class in each group is known.
- Broad applications in social science, marketing, healthcase, and computer vision.
|Designing Category-level Attributes for Discriminative Visual Recognition
- Framework and methods for designing discriminative "category-level" attributes.
- State-of-the-art results in classification with few examples, and zero-shot learning.
|More projects on vision, learning and life coming soon...|
|Weak Attributes for Large-Scale Image Retrieval
- Leveraging a large pool of weak attributes comprised of automatic classifier scores and other mid-level representations for multi-attribute (label) image retrieval.
|Active Query Sensing
- A mobile visual search system which can suggest the best query view.
- ACM Multimedia 2011 Best Paper Award.
|Adaptive Dictionary Learning for Image Retrieval|