This paper introduces Web-DRO, an unsupervised dense retrieval model, which clusters documents based on web structures and reweights the groups during contrastive training. Specifically, we first leverage web graph links and contrastively train an embedding model for clustering anchor-document pairs. Then we use Group Distributional Robust Optimization to reweight different clusters of anchor-document pairs, which guides the model to assign more weights to the group with higher contrastive loss and pay more attention to the worst case during training. Our experiments on MS MARCO and BEIR show that our model, Web-DRO, significantly improves the retrieval effectiveness in unsupervised scenarios. A comparison of clustering techniques shows that training on the web graph combining URL information reaches optimal performance on clustering. Further analysis confirms that group weights are stable and valid, indicating consistent model preferences as well as effective up-weighting of valuable groups and down-weighting of uninformative ones. The code of this paper can be obtained from //github.com/OpenMatch/Web-DRO.
相關內容
Inversion methods, such as Textual Inversion, generate personalized images by incorporating concepts of interest provided by user images. However, existing methods often suffer from overfitting issues, where the dominant presence of inverted concepts leads to the absence of other desired concepts. It stems from the fact that during inversion, the irrelevant semantics in the user images are also encoded, forcing the inverted concepts to occupy locations far from the core distribution in the embedding space. To address this issue, we propose a method that guides the inversion process towards the core distribution for compositional embeddings. Additionally, we introduce a spatial regularization approach to balance the attention on the concepts being composed. Our method is designed as a post-training approach and can be seamlessly integrated with other inversion methods. Experimental results demonstrate the effectiveness of our proposed approach in mitigating the overfitting problem and generating more diverse and balanced compositions of concepts in the synthesized images. The source code is available at //github.com/zhangxulu1996/Compositional-Inversion.
Operational consistent query answering (CQA) is a recent framework for CQA based on revised definitions of repairs, which are built by applying a sequence of operations (e.g., fact deletions) starting from an inconsistent database until we reach a database that is consistent w.r.t. the given set of constraints. It has been recently shown that there are efficient approximations for computing the percentage of repairs, as well as of sequences of operations leading to repairs, that entail a given query when we focus on primary keys, conjunctive queries, and assuming the query is fixed (i.e., in data complexity). However, it has been left open whether such approximations exist when the query is part of the input (i.e., in combined complexity). We show that this is the case when we focus on self-join-free conjunctive queries of bounded generelized hypertreewidth. We also show that it is unlikely that efficient approximation schemes exist once we give up one of the adopted syntactic restrictions, i.e., self-join-freeness or bounding the generelized hypertreewidth. Towards the desired approximation schemes, we introduce a novel counting complexity class, called SpanTL, show that each problem in SpanTL admits an efficient approximation scheme by using a recent approximability result in the context of tree automata, and then place the problems of interest in SpanTL.
This paper proposes a novel design on the wireless powered communication network (WPCN) in dynamic environments under the assistance of multiple unmanned aerial vehicles (UAVs). Unlike the existing studies, where the low-power wireless nodes (WNs) often conform to the coherent harvest-then-transmit protocol, under our newly proposed double-threshold based WN type updating rule, each WN can dynamically and repeatedly update its WN type as an E-node for non-linear energy harvesting over time slots or an I-node for transmitting data over sub-slots. To maximize the total transmission data size of all the WNs over T slots, each of the UAVs individually determines its trajectory and binary wireless energy transmission (WET) decisions over times slots and its binary wireless data collection (WDC) decisions over sub-slots, under the constraints of each UAV's limited on-board energy and each WN's node type updating rule. However, due to the UAVs' tightly-coupled trajectories with their WET and WDC decisions, as well as each WN's time-varying battery energy, this problem is difficult to solve optimally. We then propose a new multi-agent based hierarchical deep reinforcement learning (MAHDRL) framework with two tiers to solve the problem efficiently, where the soft actor critic (SAC) policy is designed in tier-1 to determine each UAV's continuous trajectory and binary WET decision over time slots, and the deep-Q learning (DQN) policy is designed in tier-2 to determine each UAV's binary WDC decisions over sub-slots under the given UAV trajectory from tier-1. Both of the SAC policy and the DQN policy are executed distributively at each UAV. Finally, extensive simulation results are provided to validate the outweighed performance of the proposed MAHDRL approach over various state-of-the-art benchmarks.
Weakly supervised 3D object detection aims to learn a 3D detector with lower annotation cost, e.g., 2D labels. Unlike prior work which still relies on few accurate 3D annotations, we propose a framework to study how to leverage constraints between 2D and 3D domains without requiring any 3D labels. Specifically, we employ visual data from three perspectives to establish connections between 2D and 3D domains. First, we design a feature-level constraint to align LiDAR and image features based on object-aware regions. Second, the output-level constraint is developed to enforce the overlap between 2D and projected 3D box estimations. Finally, the training-level constraint is utilized by producing accurate and consistent 3D pseudo-labels that align with the visual data. We conduct extensive experiments on the KITTI dataset to validate the effectiveness of the proposed three constraints. Without using any 3D labels, our method achieves favorable performance against state-of-the-art approaches and is competitive with the method that uses 500-frame 3D annotations. Code and models will be made publicly available at //github.com/kuanchihhuang/VG-W3D.
We present a general framework for preconditioning Hermitian positive definite linear systems based on the Bregman log determinant divergence. This divergence provides a measure of discrepancy between a preconditioner and a target matrix. Given an approximate factorisation of a target matrix, the proposed framework tells us how to construct a low-rank approximation of the typically indefinite factorisation error. The resulting preconditioner is therefore a sum of a Hermitian positive definite matrix given by an approximate factorisation plus a low-rank matrix. Notably, the low-rank term is not generally obtained as a truncated singular value decomposition. This framework leads to a new truncation where principal directions are not based on the magnitude of the singular values. We describe a procedure for determining these \emph{Bregman directions} and prove that preconditioners constructed in this way are minimisers of the aforementioned divergence. Finally, we demonstrate using several numerical examples how the proposed preconditioner performs in terms of convergence of the preconditioned conjugate gradient method (PCG). For the examples we consider, an incomplete Cholesky preconditioner can be greatly improved in this way, and in some cases only a modest low-rank compensation term is required to obtain a considerable improvement in convergence. We also consider matrices arising from interior point methods for linear programming that do not admit such an incomplete factorisation by default, and present a robust incomplete Cholesky preconditioner based on the proposed methodology. The results highlight that the choice of truncation is critical for ill-conditioned matrices. We show numerous examples where PCG converges to a small tolerance by using the proposed preconditioner, whereas PCG with a SVD-based preconditioner fails to do so.
This work focuses on defending against the data poisoning based backdoor attacks, which bring in serious security threats to deep neural networks (DNNs). Specifically, given a untrustworthy training dataset, we aim to filter out potential poisoned samples, \ie, poisoned sample detection (PSD). The key solution for this task is to find a discriminative metric between clean and poisoned samples, even though there is no information about the potential poisoned samples (\eg, the attack method, the poisoning ratio). In this work, we develop an innovative detection approach from the perspective of the gradient \wrt activation (\ie, activation gradient direction, AGD) of each sample in the backdoored model trained on the untrustworthy dataset. We present an interesting observation that the circular distribution of AGDs among all samples of the target class is much more dispersed than that of one clean class. Motivated by this observation, we firstly design a novel metric called Cosine similarity Variation towards Basis Transition (CVBT) to measure the circular distribution's dispersion of each class. Then, we design a simple yet effective algorithm with identifying the target class(es) using outlier detection on CVBT scores of all classes, followed by progressively filtering of poisoned samples according to the cosine similarities of AGDs between every potential sample and a few additional clean samples. Extensive experiments under various settings verify that given very few clean samples of each class, the proposed method could filter out most poisoned samples, while avoiding filtering out clean samples, verifying its effectiveness on the PSD task. Codes are available at //github.com/SCLBD/bdzoo2/blob/dev/detection_pretrain/agpd.py.
This paper focuses on parameter selection issues of kernel ridge regression (KRR). Due to special spectral properties of KRR, we find that delicate subdivision of the parameter interval shrinks the difference between two successive KRR estimates. Based on this observation, we develop an early-stopping type parameter selection strategy for KRR according to the so-called Lepskii-type principle. Theoretical verifications are presented in the framework of learning theory to show that KRR equipped with the proposed parameter selection strategy succeeds in achieving optimal learning rates and adapts to different norms, providing a new record of parameter selection for kernel methods.
Recent contrastive representation learning methods rely on estimating mutual information (MI) between multiple views of an underlying context. E.g., we can derive multiple views of a given image by applying data augmentation, or we can split a sequence into views comprising the past and future of some step in the sequence. Contrastive lower bounds on MI are easy to optimize, but have a strong underestimation bias when estimating large amounts of MI. We propose decomposing the full MI estimation problem into a sum of smaller estimation problems by splitting one of the views into progressively more informed subviews and by applying the chain rule on MI between the decomposed views. This expression contains a sum of unconditional and conditional MI terms, each measuring modest chunks of the total MI, which facilitates approximation via contrastive bounds. To maximize the sum, we formulate a contrastive lower bound on the conditional MI which can be approximated efficiently. We refer to our general approach as Decomposed Estimation of Mutual Information (DEMI). We show that DEMI can capture a larger amount of MI than standard non-decomposed contrastive bounds in a synthetic setting, and learns better representations in a vision domain and for dialogue generation.
Knowledge graph (KG) embedding encodes the entities and relations from a KG into low-dimensional vector spaces to support various applications such as KG completion, question answering, and recommender systems. In real world, knowledge graphs (KGs) are dynamic and evolve over time with addition or deletion of triples. However, most existing models focus on embedding static KGs while neglecting dynamics. To adapt to the changes in a KG, these models need to be re-trained on the whole KG with a high time cost. In this paper, to tackle the aforementioned problem, we propose a new context-aware Dynamic Knowledge Graph Embedding (DKGE) method which supports the embedding learning in an online fashion. DKGE introduces two different representations (i.e., knowledge embedding and contextual element embedding) for each entity and each relation, in the joint modeling of entities and relations as well as their contexts, by employing two attentive graph convolutional networks, a gate strategy, and translation operations. This effectively helps limit the impacts of a KG update in certain regions, not in the entire graph, so that DKGE can rapidly acquire the updated KG embedding by a proposed online learning algorithm. Furthermore, DKGE can also learn KG embedding from scratch. Experiments on the tasks of link prediction and question answering in a dynamic environment demonstrate the effectiveness and efficiency of DKGE.
The task of detecting 3D objects in point cloud has a pivotal role in many real-world applications. However, 3D object detection performance is behind that of 2D object detection due to the lack of powerful 3D feature extraction methods. In order to address this issue, we propose to build a 3D backbone network to learn rich 3D feature maps by using sparse 3D CNN operations for 3D object detection in point cloud. The 3D backbone network can inherently learn 3D features from almost raw data without compressing point cloud into multiple 2D images and generate rich feature maps for object detection. The sparse 3D CNN takes full advantages of the sparsity in the 3D point cloud to accelerate computation and save memory, which makes the 3D backbone network achievable. Empirical experiments are conducted on the KITTI benchmark and results show that the proposed method can achieve state-of-the-art performance for 3D object detection.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191