In unsupervised domain adaptation (UDA), a target-domain model is trained by the supervised knowledge from a source domain. Although UDA has recently received much attention, most existing UDA methods have ignored the alignment in label space while mostly concentrating on alignment of feature space. Even worse, they have payed less attention to the dynamic relationship between domain alignment and discrimination, leading to degenerated performance. In this work, we propose a new kind of UDA through aligning in both the feature and label spaces (DAFL), in which a dynamic weight is designed and deployed between domain alignment and discrimination enhancement according to their conditions. Specifically, the cross-domain distribution divergence is reduced by the weighted class-level feature space alignment as well as the compacted and discriminative label space alignment. Furthermore, the balancing weight between adaptation alignment and discrimination enhancement is dynamically adjusted to regularize the adversarial domain adaptation. Then, the generalization ability of the DAFL model is enhanced by adding discrepant classification with theoretical analysis. Finally, extensive experiments validate effectiveness and superiority of the proposed approach.

Multi-view learning is a machine learning paradigm that utilizes multiple feature sets or data sources to improve learning performance and generalization. However, existing multi-view learning methods often do not capture and utilize information from different views very well, especially when the relationships between views are complex and of varying quality. In this paper, we propose a novel multi-view learning framework for the multi-view classification task, called Gated Cross-Correlation Network (GCCNet), which addresses these challenges by integrating the three key operational levels in multi-view learning: representation, fusion, and decision-making. Specifically, GCCNet contains a novel component called the Multi-View Gated Information Distributor (MVGID) to enhance noise filtering and optimize the retention of critical information. In addition, GCCNet uses cross-correlation analysis to reveal dependencies and interactions between different views, as well as integrates an adaptive weighted joint decision strategy to mitigate the interference of low-quality views. Thus, GCCNet can not only comprehensively capture and utilize information from different views, but also facilitate information exchange and synergy between views, ultimately improving the overall performance of the model. Extensive experimental results on ten benchmark datasets show GCCNet's outperforms state-of-the-art methods on eight out of ten datasets, validating its effectiveness and superiority in multi-view learning.

Efficient visual detection is a crucial component in self-driving perception and lays the foundation for later planning and control stages. Deep-networks-based visual systems achieve state-of-the-art performance, but they are usually cumbersome and computationally infeasible for embedded devices (e.g., dash cams). Knowledge distillation is an effective way to derive more efficient models. However, most existing works target classification tasks and treat all instances equally. In this paper, we first present our Adaptive Instance Distillation (AID) method for self-driving visual detection. It can selectively impart the teacher's knowledge to the student by re-weighing each instance and each scale for distillation based on the teacher's loss. In addition, to enable the student to effectively digest knowledge from multiple sources, we also propose a Multi-Teacher Adaptive Instance Distillation (M-AID) method. Our M-AID helps the student to learn the best knowledge from each teacher w.r.t. certain instances and scales. Unlike previous KD methods, our M-AID adjusts the distillation weights in an instance, scale, and teacher adaptive manner. Experiments on the KITTI, COCO-Traffic, and SODA10M datasets show that our methods improve the performance of a wide variety of state-of-the-art KD methods on different detectors in self-driving scenarios. Compared to the baseline, our AID leads to an average of 2.28% and 2.98% mAP increases for single-stage and two-stage detectors, respectively. By strategically integrating knowledge from multiple teachers, our M-AID method achieves an average of 2.92% mAP improvement.

Domain adaptation (DA) aims at facilitating the target model training by leveraging knowledge from related but distribution-inconsistent source domain. Most of the previous DA works concentrate on homogeneous scenarios, where the source and target domains are assumed to share the same feature space. Nevertheless, frequently, in reality, the domains are not consistent in not only data distribution but also the representation space and feature dimensions. That is, these domains are heterogeneous. Although many works have attempted to handle such heterogeneous DA (HDA) by transforming HDA to homogeneous counterparts or performing DA jointly with domain transformation, nearly all of them just concentrate on the feature and distribution alignment across domains, neglecting the structure and classification space preservation for domains themselves. In this work, we propose a novel HDA model, namely, heterogeneous classification space alignment (HCSA), which leverages knowledge from both the source samples and model parameters to the target. In HCSA, structure preservation, distribution, and classification space alignment are implemented, jointly with feature representation by transferring both the source-domain representation and model knowledge. Moreover, we design an alternating algorithm to optimize the HCSA model with guaranteed convergence and complexity analysis. In addition, the HCSA model is further extended with deep network architecture. Finally, we experimentally evaluate the effectiveness of the proposed method by showing its superiority to the compared approaches.