Detecting and responding to novel situations in open-world environments is a key capability of human cognition. Current artificial intelligence (AI) researchers strive to develop systems that can perform in open-world environments. Novelty detection is an important ability of such AI systems. In an open-world, novelties appear in various forms and the difficulty to detect them varies. Therefore, to accurately evaluate the detection capability of AI systems, it is necessary to investigate the difficulty to detect novelties. In this paper, we propose a qualitative physics-based method to quantify the difficulty of novelty detection focusing on open-world physical domains. We apply our method in a popular physics simulation game, Angry Birds. We conduct an experiment with human players with different novelties in Angry Birds to validate our method. Results indicate that the calculated difficulty values are in line with the detection difficulty of the human players.

Automated skin lesion analysis is very crucial in clinical practice, as skin cancer is among the most common human malignancy. Existing approaches with deep learning have achieved remarkable performance on this challenging task, however, heavily relying on large-scale labelled datasets. In this paper, we present a novel active learning framework for cost-effective skin lesion analysis. The goal is to effectively select and utilize much fewer labelled samples, while the network can still achieve state-of-the-art performance. Our sample selection criteria complementarily consider both informativeness and representativeness, derived from decoupled aspects of measuring model certainty and covering sample diversity. To make wise use of the selected samples, we further design a simple yet effective strategy to aggregate intra-class images in pixel space, as a new form of data augmentation. We validate our proposed method on data of ISIC 2017 Skin Lesion Classification Challenge for two tasks. Using only up to 50% of samples, our approach can achieve state-of-the-art performances on both tasks, which are comparable or exceeding the accuracies with full-data training, and outperform other well-known active learning methods by a large margin.