Loading...
Please wait, while we are loading the content...
Similar Documents
Attributes for classifier feedback
| Content Provider | Indraprastha Institute of Information Technology, Delhi |
|---|---|
| Author | Parkash, Amar |
| Abstract | Traditional active learning allows a (machine) learner to query the (human) teacher for labels on examples it nds confusing. The teacher then provides a label for only that instance. This is quite restrictive. In this report, we propose a learning paradigm in which the learner com- municates its belief (i.e. predicted label) about the actively chosen example to the teacher. The teacher then con rms or rejects the predicted label. More importantly, if rejected, the teacher communicates an explanation for why the learners belief was wrong. This explanation allows the learner to propagate the label provided by the teacher to many unlabeled images. This allows a classi er to better learn from its mistakes, leading to accelerated discriminative learning of visual concepts even with few labeled images. In order for such communication to be feasible, it is crucial to have a language that both the human supervisor and the machine learner understand. Attributes provide precisely this channel. They are human-interpretable mid-level visual concepts shareable across categories e.g. furry, spacious, etc. We advocate the use of attributes for a supervisor to provide feedback to a Classifier and directly communicate his knowledge of the world. We employ a straightforward approach to incorporate this feedback in the Classifier, and demonstrate its power on a variety of visual recognition scenarios such as image classifi cation and annotation. This application of attributes for providing Classifiers feedback is very powerful, and has not been explored in the community. It introduces a new mode of supervision, and opens up several avenues for future research. In this work, we compare the annoyance level of the mistakes made by the Classifier in the attribute space to the mistakes made by the Classifier in gist space. |
| File Format | |
| Language | English |
| Access Restriction | Authorized |
| Subject Keyword | Attributes Classifier Feedback Visual recognition Active learning Annoyance Relative attributes Scenes Nearest neighbor |
| Content Type | Text |
| Educational Degree | Bachelor of Technology (B.Tech.) |
| Resource Type | Thesis |
| Subject | Data processing & computer science |
