Learning Multiple Dense Prediction Tasks from Partially Annotated Data
A more realistic and general setting for MTL and an architecture-agnostic algorithm to learn MTL on partially annotated data.
Research
Not All Relations are Equal: Mining Informative Labels for Scene Graph Generation
A novel model-agnostic training framework for scene graph generation based on the concept of label informativeness.
Cross-domain Few-shot Learning with Task-specific Adapters
A task-adaptation technique and systematic analysis for cross-domain few-shot classification.
Universal Representation Learning from Multiple Domains for Few-shot Classification
A universal representation learning technique for cross-domain few-shot classification.
Dataset Condensation with Gradient Matching
A training set synthesis technique for condensing a large dataset into a small set of informative synthetic samples.
Injecting Prior Knowledge into Image Caption Generation
We propose to improve the performance of the state-of-the-art image captioning models by incorporating two sources of prior knowledge.
Transfer-Based Anomaly Detection
New transfer-based techniques for detecting object-level anomalies.