IJCAI 2020 论文列表

The self-supervised learning-based depth and visual odometry (VO) estimators trained on monocular videos without ground truth have drawn significant attention recently. Prior works use photometric consistency as supervision, which is fragile under complex realistic environments due to illumination variations. More importantly, it suffers from scale inconsistency in the depth and pose estimation results. In this paper, robust geometric losses are proposed to deal with this problem. Specifically, we first align the scales of two reconstructed depth maps estimated from the adjacent image frames, and then enforce forward-backward relative pose consistency to formulate scale-consistent geometric constraints. Finally, a novel training framework is constructed to implement the proposed losses. Extensive evaluations on KITTI and Make3D datasets demonstrate that, i) by incorporating the proposed constraints as supervision, the depth estimation model can achieve state-of-the-art (SOTA) performance among the self-supervised methods, and ii) it is effective to use the proposed training framework to obtain a uniform global scale VO model.

Consensus clustering provides a framework to ensemble multiple clustering results to obtain a consensus and robust result. Most existing consensus clustering methods usually apply all data to ensemble learning, whereas ignoring the side effects caused by some difficult or unreliable instances. To tackle this problem, we propose a novel self-paced consensus clustering method to gradually involve instances from more reliable to less reliable ones into the ensemble learning. We first construct an initial bipartite graph from the multiple base clustering results, where the nodes represent the instances and clusters and the edges indicate that an instance belongs to a cluster. Then, we learn a structured bipartite graph from the initial one by self-paced learning, i.e., we automatically decide the reliability of each edge and involves the edges into graph learning in order of their reliability. At last, we obtain the final consensus clustering result from the learned bipartite graph. The extensive experimental results demonstrate the effectiveness and superiority of the proposed method.

Existing adversarial domain adaptation methods mainly consider the marginal distribution and these methods may lead to either under transfer or negative transfer. To address this problem, we present a self-adaptive re-weighted adversarial domain adaptation approach, which tries to enhance domain alignment from the perspective of conditional distribution. In order to promote positive transfer and combat negative transfer, we reduce the weight of the adversarial loss for aligned features while increasing the adversarial force for those poorly aligned measured by the conditional entropy. Additionally, triplet loss leveraging source samples and pseudo-labeled target samples is employed on the confusing domain. Such metric loss ensures the distance of the intra-class sample pairs closer than the inter-class pairs to achieve the class-level alignment. In this way, the high accurate pseudolabeled target samples and semantic alignment can be captured simultaneously in the co-training process. Our method achieved low joint error of the ideal source and target hypothesis. The expected target error can then be upper bounded following Ben-David’s theorem. Empirical evidence demonstrates that the proposed model outperforms state of the arts on standard domain adaptation datasets.

Few-shot segmentation aims at assigning a category label to each image pixel with few annotated samples. It is a challenging task since the dense prediction can only be achieved under the guidance of latent features defined by sparse annotations. Existing meta-learning based method tends to fail in generating category-specifically discriminative descriptor when the visual features extracted from support images are marginalized in embedding space. To address this issue, this paper presents an adaptive tuning framework, in which the distribution of latent features across different episodes is dynamically adjusted based on a self-segmentation scheme, augmenting category-specific descriptors for label prediction. Specifically, a novel self-supervised inner-loop is firstly devised as the base learner to extract the underlying semantic features from the support image. Then, gradient maps are calculated by back-propagating self-supervised loss through the obtained features, and leveraged as guidance for augmenting the corresponding elements in the embedding space. Finally, with the ability to continuously learn from different episodes, an optimization-based meta-learner is adopted as outer loop of our proposed framework to gradually refine the segmentation results. Extensive experiments on benchmark PASCAL-5i and COCO-20i datasets demonstrate the superiority of our proposed method over state-of-the-art.

Gait-based person re-identification (Re-ID) is valuable for safety-critical applications, and using only 3D skeleton data to extract discriminative gait features for person Re-ID is an emerging open topic. Existing methods either adopt hand-crafted features or learn gait features by traditional supervised learning paradigms. Unlike previous methods, we for the first time propose a generic gait encoding approach that can utilize unlabeled skeleton data to learn gait representations in a self-supervised manner. Specifically, we first propose to introduce self-supervision by learning to reconstruct input skeleton sequences in reverse order, which facilitates learning richer high-level semantics and better gait representations. Second, inspired by the fact that motion's continuity endows temporally adjacent skeletons with higher correlations (“locality”), we propose a locality-aware attention mechanism that encourages learning larger attention weights for temporally adjacent skeletons when reconstructing current skeleton, so as to learn locality when encoding gait. Finally, we propose Attention-based Gait Encodings (AGEs), which are built using context vectors learned by locality-aware attention, as final gait representations. AGEs are directly utilized to realize effective person Re-ID. Our approach typically improves existing skeleton-based methods by 10-20% Rank-1 accuracy, and it achieves comparable or even superior performance to multi-modal methods with extra RGB or depth information.

Evolution Strategies (ES) are a class of black-box optimization algorithms and have been widely applied to solve problems, e.g., in reinforcement learning (RL), where the true gradient is unavailable. ES estimate the gradient of an objective function with respect to the parameters by randomly sampling search directions and evaluating parameter perturbations in these directions. However, the gradient estimator of ES tends to have a high variance for high-dimensional optimization, thus requiring a large number of samples and making ES inefficient. In this paper, we propose a new ES algorithm SGES, which utilizes historical estimated gradients to construct a low-dimensional subspace for sampling search directions, and adjusts the importance of this subspace adaptively. We prove that the variance of the gradient estimator of SGES can be much smaller than that of Vanilla ES; meanwhile, its bias can be well bounded. Empirical results on benchmark black-box functions and a set of popular RL tasks exhibit the superior performance of SGES over state-of-the-art ES algorithms.

The ability to transfer knowledge to novel environments and tasks is a sensible desiderata for general learning agents. Despite the apparent promises, transfer in RL is still an open and little exploited research area. In this paper, we take a brand-new perspective about transfer: we suggest that the ability to assign credit unveils structural invariants in the tasks that can be transferred to make RL more sample-efficient. Our main contribution is SECRET, a novel approach to transfer learning for RL that uses a backward-view credit assignment mechanism based on a self-attentive architecture. Two aspects are key to its generality: it learns to assign credit as a separate offline supervised process and exclusively modifies the reward function. Consequently, it can be supplemented by transfer methods that do not modify the reward function and it can be plugged on top of any RL algorithm.

While manipulative attacks on elections have been well-studied, only recently has attention turned to attacks that account for geographic information, which are extremely common in the real world. The most well known in the media is gerrymandering, in which district border-lines are changed to increase a party's chance to win, but a different geographical manipulation involves influencing the election by selecting the location of polling places, as many people are not willing to go to any distance to vote. In this paper we initiate the study of this manipulation. We find that while it is easy to manipulate the selection of polling places on the line, it becomes difficult already on the plane or in the case of more than two candidates. Moreover, we show that for more than two candidates the problem is inapproximable. However, we find a few restricted cases on the plane where some algorithms perform well. Finally, we discuss how existing results for standard control actions hold in the geographic setting, consider additional control actions in the geographic setting, and suggest directions for future study.

Convolutional neural networks (CNNs) have achieved remarkable success in image recognition. Although the internal patterns of the input images are effectively learned by the CNNs, these patterns only constitute a small proportion of useful patterns contained in the input images. This can be attributed to the fact that the CNNs will stop learning if the learned patterns are enough to make a correct classification. Network regularization methods like dropout and SpatialDropout can ease this problem. During training, they randomly drop the features. These dropout methods, in essence, change the patterns learned by the networks, and in turn, forces the networks to learn other patterns to make the correct classification. However, the above methods have an important drawback. Randomly dropping features is generally inefficient and can introduce unnecessary noise. To tackle this problem, we propose SelectScale. Instead of randomly dropping units, SelectScale selects the important features in networks and adjusts them during training. Using SelectScale, we improve the performance of CNNs on CIFAR and ImageNet.

Besides local features, global information plays an essential role in semantic segmentation, while recent works usually fail to explicitly extract the meaningful global information and make full use of it. In this paper, we propose a SceneEncoder module to impose a scene-aware guidance to enhance the effect of global information. The module predicts a scene descriptor, which learns to represent the categories of objects existing in the scene and directly guides the point-level semantic segmentation through filtering out categories not belonging to this scene. Additionally, to alleviate segmentation noise in local region, we design a region similarity loss to propagate distinguishing features to their own neighboring points with the same label, leading to the enhancement of the distinguishing ability of point-wise features. We integrate our methods into several prevailing networks and conduct extensive experiments on benchmark datasets ScanNet and ShapeNet. Results show that our methods greatly improve the performance of baselines and achieve state-of-the-art performance.

Abstraction Sampling (AS) is a recently introduced enhancement of Importance Sampling that exploits stratification by using a notion of abstractions: groupings of similar nodes into abstract states. It was previously shown that AS performs particularly well when sampling over an AND/OR search space; however, existing schemes were limited to ``proper'' abstractions in order to ensure unbiasedness, severely hindering scalability. In this paper, we introduce AOAS, a new Abstraction Sampling scheme on AND/OR search spaces that allow more flexible use of abstractions by circumventing the properness requirement. We analyze the properties of this new algorithm and, in an extensive empirical evaluation on five benchmarks, over 480 problems, and comparing against other state of the art algorithms, illustrate AOAS's properties and show that it provides a far more powerful and competitive Abstraction Sampling framework.

Gaussian process regression networks (GPRN) are powerful Bayesian models for multi-output regression, but their inference is intractable. To address this issue, existing methods use a fully factorized structure (or a mixture of such structures) over all the outputs and latent functions for posterior approximation, which, however, can miss the strong posterior dependencies among the latent variables and hurt the inference quality. In addition, the updates of the variational parameters are inefficient and can be prohibitively expensive for a large number of outputs. To overcome these limitations, we propose a scalable variational inference algorithm for GPRN, which not only captures the abundant posterior dependencies but also is much more efficient for massive outputs. We tensorize the output space and introduce tensor/matrix-normal variational posteriors to capture the posterior correlations and to reduce the parameters. We jointly optimize all the parameters and exploit the inherent Kronecker product structure in the variational model evidence lower bound to accelerate the computation. We demonstrate the advantages of our method in several real-world applications.

Stochastic variance-reduced gradient (SVRG) is an optimization method originally designed for tackling machine learning problems with a finite sum structure. SVRG was later shown to work for policy evaluation, a problem in reinforcement learning in which one aims to estimate the value function of a given policy. SVRG makes use of gradient estimates at two scales. At the slower scale, SVRG computes a full gradient over the whole dataset, which could lead to prohibitive computation costs. In this work, we show that two variants of SVRG for policy evaluation could significantly diminish the number of gradient calculations while preserving a linear convergence speed. More importantly, our theoretical result implies that one does not need to use the entire dataset in every epoch of SVRG when it is applied to policy evaluation with linear function approximation. Our experiments demonstrate large computational savings provided by the proposed methods.

Bayesian methods have improved the interpretability and stability of neural architecture search (NAS). In this paper, we propose a novel probabilistic approach, namely Semi-Implicit Variational Dropout one-shot Neural Architecture Search (SI-VDNAS), that leverages semi-implicit variational dropout to support architecture search with variable operations and edges. SI-VDNAS achieves stable training that would not be affected by the over-selection of skip-connect operation. Experimental results demonstrate that SI-VDNAS finds a convergent architecture with only 2.7 MB parameters within 0.8 GPU-days and can achieve 2.60% top-1 error rate on CIFAR-10. The convergent architecture can obtain a top-1 error rate of 16.20% and 25.6% when transferred to CIFAR-100 and ImageNet (mobile setting).

The traditional blockchain has the shortcoming that a single-chain can only deal with one or a few specific data types. The research question of how to make blockchain be able to deal with various data types has not been well studied. In this paper, we propose a single-chain based extension model of blockchain for fintech (SEBF). In the financial environment, we design a four-layer architecture for this model. By employing the external trusted or-acle group and a financial regulator agency, a variety types of data can be effectively stored in the blockchain, such that the data type extension based on a single-chain is realized. The experimental results indicate that the proposed model can improve the efficiency of simplified payment verifi-cation.

In this paper, we focus on the problem of applying the transformer structure to video captioning effectively. The vanilla transformer is proposed for uni-modal language generation task such as machine translation. However, video captioning is a multimodal learning problem, and the video features have much redundancy between different time steps. Based on these concerns, we propose a novel method called sparse boundary-aware transformer (SBAT) to reduce the redundancy in video representation. SBAT employs boundary-aware pooling operation for scores from multihead attention and selects diverse features from different scenarios. Also, SBAT includes a local correlation scheme to compensate for the local information loss brought by sparse operation. Based on SBAT, we further propose an aligned cross-modal encoding scheme to boost the multimodal interaction. Experimental results on two benchmark datasets show that SBAT outperforms the state-of-the-art methods under most of the metrics.

The wide spread of rumors on social media has caused tremendous effects in both the online and offline world. In addition to text information, recent detection methods began to exploit the graph structure in the propagation network. However, without a rigorous design, rumors may evade such graph models using various camouflage strategies by perturbing the structured data. Our focus in this work is to develop a robust graph-based detector to identify rumors on social media from an adversarial perspective. We first build a heterogeneous information network to model the rich information among users, posts, and user comments for detection. We then propose a graph adversarial learning framework, where the attacker tries to dynamically add intentional perturbations on the graph structure to fool the detector, while the detector would learn more distinctive structure features to resist such perturbations. In this way, our model would be enhanced in both robustness and generalization. Experiments on real-world datasets demonstrate that our model achieves better results than the state-of-the-art methods.

Detecting anomalies is an important task in a wide variety of applications and domains. Deep learning methods have achieved state-of-the-art performance in anomaly detection in recent years; unsupervised methods being particularly popular. However, deep learning methods can be fragile to small perturbations in the input data. This can be exploited by an adversary to deliberately hinder model performance; an adversarial attack. This phenomena has been widely studied in the context of supervised image classification since its discovery, however such studies for an anomaly detection setting are sorely lacking. Moreover, the plethora of defense mechanisms that have been proposed are often not applicable to unsupervised anomaly detection models. In this work, we study the effect of adversarial attacks on the performance of anomaly-detecting autoencoders using real data from a Cyber physical system (CPS) testbed with intervals of controlled, physical attacks as anomalies. An adversary would attempt to disguise these points as normal through adversarial perturbations. To combat this, we propose the Approximate Projection Autoencoder (APAE), which incorporates two defenses against such attacks into a general autoencoder. One of these involves a novel technique to improve robustness under adversarial impact by optimising latent representations for better reconstruction outputs.

Most existing works in Probabilistic Simple Temporal Networks (PSTNs) base their frameworks on well-defined probability distributions. This paper addresses on PSTN Dynamic Controllability (DC) robustness measure, i.e. the execution success probability of a network under dynamic control. We consider PSTNs where the probability distributions of the contingent edges are ordinary distributed (e.g. non-parametric, non-symmetric). We introduce the concepts of dispatching protocol (DP) as well as DP-robustness, the probability of success under a predefined dynamic policy. We propose a fixed-parameter pseudo-polynomial time algorithm to compute the exact DP-robustness of any PSTN under NextFirst protocol, and apply to various PSTN datasets, including the real case of planetary exploration in the context of the Mars 2020 rover, and propose an original structural analysis.