A major direction in differentially private machine learning is differentially private fine-tuning: pretraining a model on a source of "public data" and transferring the extracted features to downstream tasks. This is an important setting because many industry deployments fine-tune publicly available feature extractors on proprietary data for downstream tasks. In this paper, we use features extracted from state-of-the-art open source models to solve benchmark tasks in computer vision and natural language processing using differentially private fine-tuning. Our key insight is that by accelerating training, we can quickly drive the model parameters to regions in parameter space where the impact of noise is minimized. In doing so, we recover the same performance as non-private fine-tuning for realistic values of epsilon in [0.01, 1.0] on benchmark image classification datasets including CIFAR100.

多传感器融合对于准确可靠的自主驾驶系统至关重要。最近的方法基于点级融合：通过相机功能增强激光雷达点云。但是，摄像头投影抛弃了相机功能的语义密度，阻碍了此类方法的有效性，尤其是对于面向语义的任务（例如3D场景分割）。在本文中，我们用BevFusion打破了这个根深蒂固的惯例，这是一个有效且通用的多任务多任务融合框架。它统一了共享鸟类视图（BEV）表示空间中的多模式特征，该空间很好地保留了几何信息和语义信息。为了实现这一目标，我们通过优化的BEV池进行诊断和提高视图转换中的钥匙效率瓶颈，从而将延迟降低了40倍以上。 BevFusion从根本上是任务不合时宜的，并且无缝支持不同的3D感知任务，几乎没有建筑变化。它在Nuscenes上建立了新的最新技术，在3D对象检测上获得了1.3％的MAP和NDS，而BEV MAP分段中的MIOU高13.6％，计算成本较低1.9倍。可以在https://github.com/mit-han-lab/bevfusion上获得复制我们结果的代码。

图形神经网络（GNNS）在各种现实世界应用中取得了有希望的性能。然而，最近的研究表明，GNN易受对抗性发作的影响。在本文中，我们研究了关于图表 - 图 - 图注射攻击（GIA）的最近引入的现实攻击情景。在GIA场景中，对手无法修改输入图的现有链路结构和节点属性，而是通过将逆势节点注入到它中来执行攻击。我们对GIA环境下GNN的拓扑脆弱性分析，基于该拓扑结构，我们提出了用于有效注射攻击的拓扑缺陷图注射攻击（TDGIA）。 TDGIA首先介绍了拓扑有缺陷的边缘选择策略，可以选择与注入的原始节点连接。然后，它设计平滑功能优化目标，以生成注入节点的功能。大规模数据集的广泛实验表明，TDGIA可以一致而明显优于攻击数十个防御GNN模型中的各种攻击基线。值得注意的是，来自TDGIA的目标GNNS上的性能下降比KDD-CUP 2020上的数百个提交所带来的最佳攻击解决方案所带来的损坏多于两倍。

As one of the most important psychic stress reactions, micro-expressions (MEs), are spontaneous and transient facial expressions that can reveal the genuine emotions of human beings. Thus, recognizing MEs (MER) automatically is becoming increasingly crucial in the field of affective computing, and provides essential technical support in lie detection, psychological analysis and other areas. However, the lack of abundant ME data seriously restricts the development of cutting-edge data-driven MER models. Despite the recent efforts of several spontaneous ME datasets to alleviate this problem, it is still a tiny amount of work. To solve the problem of ME data hunger, we construct a dynamic spontaneous ME dataset with the largest current ME data scale, called DFME (Dynamic Facial Micro-expressions), which includes 7,526 well-labeled ME videos induced by 671 participants and annotated by more than 20 annotators throughout three years. Afterwards, we adopt four classical spatiotemporal feature learning models on DFME to perform MER experiments to objectively verify the validity of DFME dataset. In addition, we explore different solutions to the class imbalance and key-frame sequence sampling problems in dynamic MER respectively on DFME, so as to provide a valuable reference for future research. The comprehensive experimental results show that our DFME dataset can facilitate the research of automatic MER, and provide a new benchmark for MER. DFME will be published via https://mea-lab-421.github.io.

Reading comprehension of legal text can be a particularly challenging task due to the length and complexity of legal clauses and a shortage of expert-annotated datasets. To address this challenge, we introduce the Merger Agreement Understanding Dataset (MAUD), an expert-annotated reading comprehension dataset based on the American Bar Association's 2021 Public Target Deal Points Study, with over 39,000 examples and over 47,000 total annotations. Our fine-tuned Transformer baselines show promising results, with models performing well above random on most questions. However, on a large subset of questions, there is still room for significant improvement. As the only expert-annotated merger agreement dataset, MAUD is valuable as a benchmark for both the legal profession and the NLP community.

An increasing number of public datasets have shown a marked clinical impact on assessing anatomical structures. However, each of the datasets is small, partially labeled, and rarely investigates severe tumor subjects. Moreover, current models are limited to segmenting specific organs/tumors, which can not be extended to novel domains and classes. To tackle these limitations, we introduce embedding learned from Contrastive Language-Image Pre-training (CLIP) to segmentation models, dubbed the CLIP-Driven Universal Model. The Universal Model can better segment 25 organs and 6 types of tumors by exploiting the semantic relationship between abdominal structures. The model is developed from an assembly of 14 datasets with 3,410 CT scans and evaluated on 6,162 external CT scans from 3 datasets. We rank first on the public leaderboard of the Medical Segmentation Decathlon (MSD) and achieve the state-of-the-art results on Beyond The Cranial Vault (BTCV). Compared with dataset-specific models, the Universal Model is computationally more efficient (6x faster), generalizes better to CT scans from varying sites, and shows stronger transfer learning performance on novel tasks. The design of CLIP embedding enables the Universal Model to be easily extended to new classes without catastrophically forgetting the previously learned classes.

In recent years, the Transformer architecture has shown its superiority in the video-based person re-identification task. Inspired by video representation learning, these methods mainly focus on designing modules to extract informative spatial and temporal features. However, they are still limited in extracting local attributes and global identity information, which are critical for the person re-identification task. In this paper, we propose a novel Multi-Stage Spatial-Temporal Aggregation Transformer (MSTAT) with two novel designed proxy embedding modules to address the above issue. Specifically, MSTAT consists of three stages to encode the attribute-associated, the identity-associated, and the attribute-identity-associated information from the video clips, respectively, achieving the holistic perception of the input person. We combine the outputs of all the stages for the final identification. In practice, to save the computational cost, the Spatial-Temporal Aggregation (STA) modules are first adopted in each stage to conduct the self-attention operations along the spatial and temporal dimensions separately. We further introduce the Attribute-Aware and Identity-Aware Proxy embedding modules (AAP and IAP) to extract the informative and discriminative feature representations at different stages. All of them are realized by employing newly designed self-attention operations with specific meanings. Moreover, temporal patch shuffling is also introduced to further improve the robustness of the model. Extensive experimental results demonstrate the effectiveness of the proposed modules in extracting the informative and discriminative information from the videos, and illustrate the MSTAT can achieve state-of-the-art accuracies on various standard benchmarks.

Neural models with an encoder-decoder framework provide a feasible solution to Question Generation (QG). However, after analyzing the model vocabulary we find that current models (both RNN-based and pre-training based) have more than 23\% inflected forms. As a result, the encoder will generate separate embeddings for the inflected forms, leading to a waste of training data and parameters. Even worse, in decoding these models are vulnerable to irrelevant noise and they suffer from high computational costs. In this paper, we propose an approach to enhance the performance of QG by fusing word transformation. Firstly, we identify the inflected forms of words from the input of encoder, and replace them with the root words, letting the encoder pay more attention to the repetitive root words. Secondly, we propose to adapt QG as a combination of the following actions in the encode-decoder framework: generating a question word, copying a word from the source sequence or generating a word transformation type. Such extension can greatly decrease the size of predicted words in the decoder as well as noise. We apply our approach to a typical RNN-based model and \textsc{UniLM} to get the improved versions. We conduct extensive experiments on SQuAD and MS MARCO datasets. The experimental results show that the improved versions can significantly outperform the corresponding baselines in terms of BLEU, ROUGE-L and METEOR as well as time cost.

In this paper, we develop an efficient multi-scale network to predict action classes in partial videos in an end-to-end manner. Unlike most existing methods with offline feature generation, our method directly takes frames as input and further models motion evolution on two different temporal scales.Therefore, we solve the complexity problems of the two stages of modeling and the problem of insufficient temporal and spatial information of a single scale. Our proposed End-to-End MultiScale Network (E2EMSNet) is composed of two scales which are named segment scale and observed global scale. The segment scale leverages temporal difference over consecutive frames for finer motion patterns by supplying 2D convolutions. For observed global scale, a Long Short-Term Memory (LSTM) is incorporated to capture motion features of observed frames. Our model provides a simple and efficient modeling framework with a small computational cost. Our E2EMSNet is evaluated on three challenging datasets: BIT, HMDB51, and UCF101. The extensive experiments demonstrate the effectiveness of our method for action prediction in videos.

The ability to jointly learn from multiple modalities, such as text, audio, and visual data, is a defining feature of intelligent systems. While there have been promising advances in designing neural networks to harness multimodal data, the enormous success of data augmentation currently remains limited to single-modality tasks like image classification. Indeed, it is particularly difficult to augment each modality while preserving the overall semantic structure of the data; for example, a caption may no longer be a good description of an image after standard augmentations have been applied, such as translation. Moreover, it is challenging to specify reasonable transformations that are not tailored to a particular modality. In this paper, we introduce LeMDA, Learning Multimodal Data Augmentation, an easy-to-use method that automatically learns to jointly augment multimodal data in feature space, with no constraints on the identities of the modalities or the relationship between modalities. We show that LeMDA can (1) profoundly improve the performance of multimodal deep learning architectures, (2) apply to combinations of modalities that have not been previously considered, and (3) achieve state-of-the-art results on a wide range of applications comprised of image, text, and tabular data.