Search

Open-World Object Detection (OWOD) extends object detection problem to a realistic and dynamic scenario, where a detection model is required to be capable of detecting both known and unknown objects and incrementally learning newly introduced knowledge. Current OWOD models, such as ORE and OW-DETR, focus on pseudo-labeling regions with high objectness scores as unknowns, whose performance relies heavily on the supervision of known objects. While they can detect the unknowns that exhibit similar features to the known objects, they suffer from a severe label bias problem that they tend to detect all regions (including unknown object regions) that are dissimilar to the known objects as part of the background. To eliminate the label bias, this paper proposes a novel approach that learns an unsupervised discriminative model to recognize true unknown objects from raw pseudo labels generated by unsupervised region proposal methods. The resulting model can be further refined by a classification-free self-training method which iteratively extends pseudo unknown objects to the unlabeled regions. Experimental results show that our method 1) significantly outperforms the prior SOTA in detecting unknown objects while maintaining competitive performance of detecting known object classes on the MS COCO dataset, and 2) achieves better generalization ability on the LVIS and Objects365 datasets.

Temporal knowledge graphs, representing the dynamic relationships and interactions between entities over time, have been identified as a promising approach for event forecasting. However, a limitation of most temporal knowledge graph reasoning methods is their heavy reliance on the recurrence or periodicity of events, which brings challenges to inferring future events related to entities that lack historical interaction. In fact, the current state of affairs is often the result of a combination of historical information and underlying factors that are not directly observable. To this end, we investigate the limits of historical information for temporal knowledge graph extrapolation and propose a new event forecasting model called Contrastive Event Network (CENET) based on a novel training framework of historical contrastive learning. CENET learns both the historical and non-historical dependency to distinguish the most potential entities that best match the given query. Simultaneously, by launching contrastive learning, it trains representations of queries to probe whether the current moment is more dependent on historical or non-historical events. These representations further help train a binary classifier, whose output is a boolean mask, indicating the related entities in the search space. During the inference process, CENET employs a mask-based strategy to generate the final results. We evaluate our proposed model on five benchmark graphs. The results demonstrate that CENET significantly outperforms all existing methods in most metrics, achieving at least 8.3% relative improvement of Hits@1 over previous state-of-the-art baselines on event-based datasets., Comment: Extended version of AAAI paper arXiv:2211.10904

The use of nanophotonics for optical manipulation has continuously attracted interest in both fundamental research and practical applications, due to its significantly enhanced capabilities at the nanoscale. In this work, we showed that plasmonic particles can be trapped at off-axis location in Gaussian beams assisted by surface plasmon resonance. The off-axis displacement can be tuned at the sub-wavelength scale by the incident light beams. Based on these, we propose that a superfast orbital rotation of particles in continuous-wave laser beam can be realized in tightly focused circularly polarized Gaussian beams. The rotation has a tunable orbital radius at the sub-wavelength scale and a superfast rotation speed (more than 10^4 r/s in water under common laboratory conditions). Our work will aid in the development of optically driven nanomachines, and find applications in micro/nano-rheology, micro-fluid mechanics, and biological research at the nanoscale., Comment: 5 pages, 5 figures

Particles trapped by optical tweezers, behaving as mechanical oscillators in an optomechanical system, have found tremendous applications in various disciplines and are still arousing research interest in frontier and fundamental physics. These optically trapped oscillators provide compact particle confinement and strong oscillator stiffness. But these features are limited by the size of the focused light spot of a laser beam, which is typically restricted by the optical diffraction limit. Here, we propose to build an optical potential well with fine features assisted by the nonlinearity of the particle material, which is independent of the optical diffraction limit. We show that the potential well shape can have super-oscillation-like features and a Fano-resonance-like phenomenon, and the width of the optical trap is far below the diffraction limit. A particle with nonlinearity trapped by an ordinary optical beam provides a new platform with a sub-diffraction potential well and can have applications in high-accuracy optical manipulation and high-precision metrology., Comment: 4 pages, 5 figures

Zero-Shot Learning (ZSL) focuses on classifying samples of unseen classes with only their side semantic information presented during training. It cannot handle real-life, open-world scenarios where there are test samples of unknown classes for which neither samples (e.g., images) nor their side semantic information is known during training. Open-Set Recognition (OSR) is dedicated to addressing the unknown class issue, but existing OSR methods are not designed to model the semantic information of the unseen classes. To tackle this combined ZSL and OSR problem, we consider the case of "Zero-Shot Open-Set Recognition" (ZS-OSR), where a model is trained under the ZSL setting but it is required to accurately classify samples from the unseen classes while being able to reject samples from the unknown classes during inference. We perform large experiments on combining existing state-of-the-art ZSL and OSR models for the ZS-OSR task on four widely used datasets adapted from the ZSL task, and reveal that ZS-OSR is a non-trivial task as the simply combined solutions perform badly in distinguishing the unseen-class and unknown-class samples. We further introduce a novel approach specifically designed for ZS-OSR, in which our model learns to generate adversarial semantic embeddings of the unknown classes to train an unknowns-informed ZS-OSR classifier. Extensive empirical results show that our method 1) substantially outperforms the combined solutions in detecting the unknown classes while retaining the classification accuracy on the unseen classes and 2) achieves similar superiority under generalized ZS-OSR settings.

Humans interact and cooperate in structured societies, which are often represented by complex networks. Previous explorations mainly focus on pairwise and static network structures, where each link connects two nodes permanently and never changes. However, empirical human collective interactions go beyond static time-invariant one-to-one relationships. Recently, researchers have made vital progress in modelling higher-order interactions using hypernetworks, where a link may connect more than two individuals. Here, we study collective cooperation through temporal hypernetworks, capturing the time-varying higher-order interactions in human societies. We find that temporal hypernetworks may promote cooperation compared with their static counterparts, and the traditional static pairwise network may underestimate the positive effect of local interactions on fostering cooperators. Moreover, temporal hypernetworks with sparse components and higher-order interactions can facilitate cooperation. Surprisingly, we report that when the scale of interactions is of the same order as population size, moderately small hyperlink sizes best facilitate cooperation. Our findings underscore the significant impact of real-world temporal higher-order interactions on the evolution of cooperation., Comment: 14 pages, 6 figures