PromptKG: A Prompt Learning Framework for Knowledge Graph Representation Learning and Application

Knowledge Graphs (KGs) often have two characteristics: heterogeneous graph structure and text-rich entity/relation information. KG representation models should consider graph structures and text semantics, but no comprehensive open-sourced framework is mainly designed for KG regarding informative text description. In this paper, we present PromptKG, a prompt learning framework for KG representation learning and application that equips the cutting-edge text-based methods, integrates a new prompt learning model and supports various tasks (e.g., knowledge graph completion, question answering, recommendation, and knowledge probing). PromptKG is publicly open-sourced at https://github.com/zjunlp/PromptKG with long-term technical support.

Slimmable Networks for Contrastive Self-supervised Learning

Self-supervised learning makes great progress in large model pre-training but suffers in training small models. Previous solutions to this problem mainly rely on knowledge distillation and indeed have a two-stage learning procedure: first train a large teacher model, then distill it to improve the generalization ability of small ones. In this work, we present a new one-stage solution to obtain pre-trained small models without extra teachers: slimmable networks for contrastive self-supervised learning (\emph{SlimCLR}). A slimmable network contains a full network and several weight-sharing sub-networks. We can pre-train for only one time and obtain various networks including small ones with low computation costs. However, in self-supervised cases, the interference between weight-sharing networks leads to severe performance degradation. One evidence of the interference is \emph{gradient imbalance}: a small proportion of parameters produces dominant gradients during backpropagation, and the main parameters may not be fully optimized. The divergence in gradient directions of various networks may also cause interference between networks. To overcome these problems, we make the main parameters produce dominant gradients and provide consistent guidance for sub-networks via three techniques: slow start training of sub-networks, online distillation, and loss re-weighting according to model sizes. Besides, a switchable linear probe layer is applied during linear evaluation to avoid the interference of weight-sharing linear layers. We instantiate SlimCLR with typical contrastive learning frameworks and achieve better performance than previous arts with fewer parameters and FLOPs.

Generalizable Memory-driven Transformer for Multivariate Long Sequence Time-series Forecasting

Multivariate long sequence time-series forecasting (M-LSTF) is a practical but challenging problem. Unlike traditional timer-series forecasting tasks, M-LSTF tasks are more challenging from two aspects: 1) M-LSTF models need to learn time-series patterns both within and between multiple time features; 2) Under the rolling forecasting setting, the similarity between two consecutive training samples increases with the increasing prediction length, which makes models more prone to overfitting. In this paper, we propose a generalizable memory-driven Transformer to target M-LSTF problems. Specifically, we first propose a global-level memory component to drive the forecasting procedure by integrating multiple time-series features. In addition, we adopt a progressive fashion to train our model to increase its generalizability, in which we gradually introduce Bernoulli noises to training samples. Extensive experiments have been performed on five different datasets across multiple fields. Experimental results demonstrate that our approach can be seamlessly plugged into varying Transformer-based models to improve their performances up to roughly 30%. Particularly, this is the first work to specifically focus on the M-LSTF tasks to the best of our knowledge.

Jointly Harnessing Prior Structures and Temporal Consistency for Sign Language Video Generation

Sign language is the window for people differently-abled to express their feelings as well as emotions. However, it remains challenging for people to learn sign language in a short time. To address this real-world challenge, in this work, we study the motion transfer system, which can transfer the user photo to the sign language video of specific words. In particular, the appearance content of the output video comes from the provided user image, while the motion of the video is extracted from the specified tutorial video. We observe two primary limitations in adopting the state-of-the-art motion transfer methods to sign language generation:(1) Existing motion transfer works ignore the prior geometrical knowledge of the human body. (2) The previous image animation methods only take image pairs as input in the training stage, which could not fully exploit the temporal information within videos. In an attempt to address the above-mentioned limitations, we propose Structure-aware Temporal Consistency Network (STCNet) to jointly optimize the prior structure of human with the temporal consistency for sign language video generation. There are two main contributions in this paper. (1) We harness a fine-grained skeleton detector to provide prior knowledge of the body keypoints. In this way, we ensure the keypoint movement in a valid range and make the model become more explainable and robust. (2) We introduce two cycle-consistency losses, i.e., short-term cycle loss and long-term cycle loss, which are conducted to assure the continuity of the generated video. We optimize the two losses and keypoint detector network in an end-to-end manner.

ReLER@ZJU-Alibaba Submission to the Ego4D Natural Language Queries Challenge 2022

In this report, we present the ReLER@ZJU-Alibaba submission to the Ego4D Natural Language Queries (NLQ) Challenge in CVPR 2022. Given a video clip and a text query, the goal of this challenge is to locate a temporal moment of the video clip where the answer to the query can be obtained. To tackle this task, we propose a multi-scale cross-modal transformer and a video frame-level contrastive loss to fully uncover the correlation between language queries and video clips. Besides, we propose two data augmentation strategies to increase the diversity of training samples. The experimental results demonstrate the effectiveness of our method. The final submission ranked first on the leaderboard.

CenterCLIP: Token Clustering for Efficient Text-Video Retrieval

Recently, large-scale pre-training methods like CLIP have made great progress in multi-modal research such as text-video retrieval. In CLIP, transformers are vital for modeling complex multi-modal relations. However, in the vision transformer of CLIP, the essential visual tokenization process, which produces discrete visual token sequences, generates many homogeneous tokens due to the redundancy nature of consecutive and similar frames in videos. This significantly increases computation costs and hinders the deployment of video retrieval models in web applications. In this paper, to reduce the number of redundant video tokens, we design a multi-segment token clustering algorithm to find the most representative tokens and drop the non-essential ones. As the frame redundancy occurs mostly in consecutive frames, we divide videos into multiple segments and conduct segment-level clustering. Center tokens from each segment are later concatenated into a new sequence, while their original spatial-temporal relations are well maintained. We instantiate two clustering algorithms to efficiently find deterministic medoids and iteratively partition groups in high dimensional space. Through this token clustering and center selection procedure, we successfully reduce computation costs by removing redundant visual tokens. This method further enhances segment-level semantic alignment between video and text representations, enforcing the spatio-temporal interactions of tokens from within-segment frames. Our method, coined as CenterCLIP, surpasses existing state-of-the-art by a large margin on typical text-video benchmarks, while reducing the training memory cost by 35\% and accelerating the inference speed by 14\% at the best case. The code is available at \href{{https://github.com/mzhaoshuai/CenterCLIP}}{{https://github.com/mzhaoshuai/CenterCLIP}}.

Associating Objects with Scalable Transformers for Video Object Segmentation

This paper investigates how to realize better and more efficient embedding learning to tackle the semi-supervised video object segmentation under challenging multi-object scenarios. The state-of-the-art methods learn to decode features with a single positive object and thus have to match and segment each target separately under multi-object scenarios, consuming multiple times computation resources. To solve the problem, we propose an Associating Objects with Transformers (AOT) approach to match and decode multiple objects jointly and collaboratively. In detail, AOT employs an identification mechanism to associate multiple targets into the same high-dimensional embedding space. Thus, we can simultaneously process multiple objects' matching and segmentation decoding as efficiently as processing a single object. To sufficiently model multi-object association, a Long Short-Term Transformer (LSTT) is devised to construct hierarchical matching and propagation. Based on AOT, we further propose a more flexible and robust framework, Associating Objects with Scalable Transformers (AOST), in which a scalable version of LSTT is designed to enable run-time adaptation of accuracy-efficiency trade-offs. Besides, AOST introduces a better layer-wise manner to couple identification and vision embeddings. We conduct extensive experiments on multi-object and single-object benchmarks to examine AOT series frameworks. Compared to the state-of-the-art competitors, our methods can maintain times of run-time efficiency with superior performance. Notably, we achieve new state-of-the-art performance on three popular benchmarks, i.e., YouTube-VOS (86.5%), DAVIS 2017 Val/Test (87.0%/84.7%), and DAVIS 2016 (93.0%). Project page: https://github.com/z-x-yang/AOT.

Multi-robot Cooperative Pursuit via Potential Field-Enhanced Reinforcement Learning

It is of great challenge, though promising, to coordinate collective robots for hunting an evader in a decentralized manner purely in light of local observations. In this paper, this challenge is addressed by a novel hybrid cooperative pursuit algorithm that combines reinforcement learning with the artificial potential field method. In the proposed algorithm, decentralized deep reinforcement learning is employed to learn cooperative pursuit policies that are adaptive to dynamic environments. The artificial potential field method is integrated into the learning process as predefined rules to improve the data efficiency and generalization ability. It is shown by numerical simulations that the proposed hybrid design outperforms the pursuit policies either learned from vanilla reinforcement learning or designed by the potential field method. Furthermore, experiments are conducted by transferring the learned pursuit policies into real-world mobile robots. Experimental results demonstrate the feasibility and potential of the proposed algorithm in learning multiple cooperative pursuit strategies.

Action Keypoint Network for Efficient Video Recognition

Reducing redundancy is crucial for improving the efficiency of video recognition models. An effective approach is to select informative content from the holistic video, yielding a popular family of dynamic video recognition methods. However, existing dynamic methods focus on either temporal or spatial selection independently while neglecting a reality that the redundancies are usually spatial and temporal, simultaneously. Moreover, their selected content is usually cropped with fixed shapes, while the realistic distribution of informative content can be much more diverse. With these two insights, this paper proposes to integrate temporal and spatial selection into an Action Keypoint Network (AK-Net). From different frames and positions, AK-Net selects some informative points scattered in arbitrary-shaped regions as a set of action keypoints and then transforms the video recognition into point cloud classification. AK-Net has two steps, i.e., the keypoint selection and the point cloud classification. First, it inputs the video into a baseline network and outputs a feature map from an intermediate layer. We view each pixel on this feature map as a spatial-temporal point and select some informative keypoints using self-attention. Second, AK-Net devises a ranking criterion to arrange the keypoints into an ordered 1D sequence. Consequentially, AK-Net brings two-fold benefits for efficiency: The keypoint selection step collects informative content within arbitrary shapes and increases the efficiency for modeling spatial-temporal dependencies, while the point cloud classification step further reduces the computational cost by compacting the convolutional kernels. Experimental results show that AK-Net can consistently improve the efficiency and performance of baseline methods on several video recognition benchmarks.

Point Cloud Pre-training by Mixing and Disentangling

The annotation for large-scale point clouds is still time-consuming and unavailable for many real-world tasks. Point cloud pre-training is one potential solution for obtaining a scalable model for fast adaptation. Therefore, in this paper, we investigate a new self-supervised learning approach, called Mixing and Disentangling (MD), for point cloud pre-training. As the name implies, we explore how to separate the original point cloud from the mixed point cloud, and leverage this challenging task as a pretext optimization objective for model training. Considering the limited training data in the original dataset, which is much less than prevailing ImageNet, the mixing process can efficiently generate more high-quality samples. We build one baseline network to verify our intuition, which simply contains two modules, encoder and decoder. Given a mixed point cloud, the encoder is first pre-trained to extract the semantic embedding. Then an instance-adaptive decoder is harnessed to disentangle the point clouds according to the embedding. Albeit simple, the encoder is inherently able to capture the point cloud keypoints after training and can be fast adapted to downstream tasks including classification and segmentation by the pre-training and fine-tuning paradigm. Extensive experiments on two datasets show that the encoder + ours (MD) significantly surpasses that of the encoder trained from scratch and converges quickly. In ablation studies, we further study the effect of each component and discuss the advantages of the proposed self-supervised learning strategy. We hope this self-supervised learning attempt on point clouds can pave the way for reducing the deeply-learned model dependence on large-scale labeled data and saving a lot of annotation costs in the future.