Open-vocabulary object detection (OVOD) aims to recognize novel objects whose categories are not included in the training set. In order to classify these unseen classes during training, many OVOD frameworks leverage the zero-shot capability of largely pretrained vision and language models, such as CLIP. To further improve generalization on the unseen novel classes, several approaches proposed to additionally train with pseudo region labeling on the external data sources that contain a substantial number of novel category labels beyond the existing training data. Albeit its simplicity, these pseudo-labeling methods still exhibit limited improvement with regard to the truly unseen novel classes that were not pseudo-labeled. In this paper, we present a novel, yet simple technique that helps generalization on the overall distribution of novel classes. Inspired by our observation that numerous novel classes reside within the convex hull constructed by the base (seen) classes in the CLIP embedding space, we propose to synthesize proxy-novel classes approximating novel classes via linear mixup between a pair of base classes. By training our detector with these synthetic proxy-novel classes, we effectively explore the embedding space of novel classes. The experimental results on various OVOD benchmarks such as LVIS and COCO demonstrate superior performance on novel classes compared to the other state-of-the-art methods. Code is available at https://github.com/clovaai/ProxyDet.

4D medical images, which represent 3D images with temporal information, are crucial in clinical practice for capturing dynamic changes and monitoring long-term dis-ease progression. However, acquiring 4D medical images poses challenges due to factors such as radiation expo-sure and imaging duration, necessitating a balance be-tween achieving high temporal resolution and minimizing adverse effects. Given these circumstances, not only is data acquisition challenging, but increasing the frame rate for each dataset also proves difficult. To address this challenge, this paper proposes a simple yet effective Unsupervised Volumetric Interpolation framework, UVI-Net. This frame-work facilitates temporal interpolation without the need for any intermediate frames, distinguishing it from the major-ity of other existing unsupervised methods. Experiments on benchmark datasets demonstrate significant improvements across diverse evaluation metrics compared to unsuper-vised and supervised baselines. Remarkably, our approach achieves this superior performance even when trained with a dataset as small as one, highlighting its exceptional ro-bustness and efficiency in scenarios with sparse supervision. This positions UVI-Net as a compelling alternative for 4D medical imaging, particularly in settings where data availability is limited. The code is available at UVI-Net.

Tabular data in the wild are frequently afflicted with class-imbalance, biasing machine learning model predictions towards major classes. A data-centric solution to this problem is oversampling - where the classes are balanced by adding synthetic minority samples via generative methods. However, although tabular generative models are capable of generating synthetic samples under a balanced distribution, their integrity suffers when the number of minority samples is low. To this end, pre-trained generative language models with rich prior knowledge are a fitting candidate for the task at hand. Nevertheless, an oversampling strategy tailored for tabular data that utilizes the extensive capabilities of such language models is yet to emerge. In this paper, we propose a novel oversampling framework for tabular data to channel the abilities of generative language models. By leveraging its conditional sampling capabilities, we synthesize minority samples by progressively masking the important features of the majority class samples and imputing them towards the minority distribution. To reduce the inclusion of imperfectly converted samples, we utilize the power of the language model itself to self-authenticate the labels of the samples generated by itself, sifting out ill-converted samples. Extensive experiments on a variety of datasets and imbalance ratios reveal that the proposed method successfully generates reliable minority samples to boost the performance of machine learning classifiers, even under heavy imbalance ratios.

Neural Architecture Search (NAS) aims to automatically excavate the optimal network architecture with superior test performance. Recent neural architecture search (NAS) approaches rely on validation loss or accuracy to find the superior network for the target data. In this paper, we investigate a new neural architecture search measure for excavating architectures with better generalization. We demonstrate that the flatness of the loss surface can be a promising proxy for predicting the generalization capability of neural network architectures. We evaluate our proposed method on various search spaces, showing similar or even better performance compared to the state-of-the-art NAS methods. Notably, the resultant architecture found by flatness measure generalizes robustly to various shifts in data distribution (e.g. ImageNet-V2,-A,-O), as well as various tasks such as object detection and semantic segmentation.

A computationally expensive and memory intensive neural network lies behind the recent success of language representation learning.Knowledge distillation, a major technique for deploying such a vast language model in resource-scarce environments, transfers the knowledge on individual word representations learned without restrictions.In this paper, inspired by the recent observations that language representations are relatively positioned and have more semantic knowledge as a whole, we present a new knowledge distillation objective for language representation learning that transfers the contextual knowledge via two types of relationships across representations: Word Relation and Layer Transforming Relation.Unlike other recent distillation techniques for the language models, our contextual distillation does not have any restrictions on architectural changes between teacher and student.We validate the effectiveness of our method on challenging benchmarks of language understanding tasks, not only in architectures of various sizes, but also in combination with DynaBERT, the recently proposed adaptive size pruning method.

Neural networks embedded in safety-sensitive applications such as self-driving cars and wearable health monitors rely on two important techniques: input attribution for hindsight analysis and network compression to reduce its size for edge-computing. In this paper, we show that these seemingly unrelated techniques conflict with each other as network compression deforms the produced attributions, which could lead to dire consequences for mission-critical applications. This phenomenon arises due to the fact that conventional network compression methods only preserve the predictions of the network while ignoring the quality of the attributions. To combat the attribution inconsistency problem, we present a framework that can preserve the attributions while compressing a network. By employing the Weighted Collapsed Attribution Matching regularizer, we match the attribution maps of the network being compressed to its pre-compression former self. We demonstrate the effectiveness of our algorithm both quantitatively and qualitatively on diverse compression methods.

Visual-semantic embedding enables various tasks such as image-text retrieval, image captioning, and visual question answering. The key to successful visual-semantic embedding is to express visual and textual data properly by accounting for their intricate relationship. While previous studies have achieved much advance by encoding the visual and textual data into a joint space where similar concepts are closely located, they often represent data by a single vector ignoring the presence of multiple important components in an image or text. Thus, in addition to the joint embedding space, we propose a novel multi-head self-attention network to capture various components of visual and textual data by attending to important parts in data. Our approach achieves the new state-of-the-art results in image-text retrieval tasks on MS-COCO and Flicker30K datasets. Through the visualization of the attention maps that capture distinct semantic components at multiple positions in the image and the text, we demonstrate that our method achieves an effective and interpretable visual-semantic joint space.