IJCAI 2020 论文列表

Zero-Shot Sketch-Based Image Retrieval (ZS-SBIR) is a specific cross-modal retrieval task for searching natural images given free-hand sketches under the zero-shot scenario. Most existing methods solve this problem by simultaneously projecting visual features and semantic supervision into a low-dimensional common space for efficient retrieval. However, such low-dimensional projection destroys the completeness of semantic knowledge in original semantic space, so that it is unable to transfer useful knowledge well when learning semantic features from different modalities. Moreover, the domain information and semantic information are entangled in visual features, which is not conducive for cross-modal matching since it will hinder the reduction of domain gap between sketch and image. In this paper, we propose a Progressive Domain-independent Feature Decomposition (PDFD) network for ZS-SBIR. Specifically, with the supervision of original semantic knowledge, PDFD decomposes visual features into domain features and semantic ones, and then the semantic features are projected into common space as retrieval features for ZS-SBIR. The progressive projection strategy maintains strong semantic supervision. Besides, to guarantee the retrieval features to capture clean and complete semantic information, the cross-reconstruction loss is introduced to encourage that any combinations of retrieval features and domain features can reconstruct the visual features. Extensive experiments demonstrate the superiority of our PDFD over state-of-the-art competitors.

Modeling object representations derived from perceptual observations, in a way that is also semantically meaningful for humans as well as autonomous agents, is a prerequisite for joint human-agent understanding of the world. A practical approach that aims to model such representations is perceptual anchoring, which handles the problem of mapping sub-symbolic sensor data to symbols and maintains these mappings over time. In this paper, we present ProbAnch, a modular data-driven anchoring framework, whose implementation requires a variety of well-orchestrated components, including a probabilistic reasoning system.

Label distribution learning has attracted more and more attention in view of its more generalized ability to express the label ambiguity. However, it is much more expensive to obtain the label distribution information of the data rather than the logical labels. Thus, label enhancement is proposed to recover the label distributions from the logical labels. In this paper, we propose a novel label enhancement method by using privileged information. We first apply a multi-label learning model to implicitly capture the complex structural information between instances and generate the privileged information. Second, we adopt LUPI (learning with privileged information) paradigm to utilize the privileged information and employ RSVM+ as the prediction model. Finally, comparison experiments on 12 datasets demonstrate that our proposal can better fit the ground-truth label distributions.

Tractable Deep Probabilistic Models (TPMs) are generative models based on arithmetic circuits that allow for exact marginal inference in linear time. These models have obtained promising results in several machine learning tasks. Like many other models, TPMs can produce over-confident incorrect inferences, especially on regions with small statistical support. In this work, we will develop efficient estimators of the predictive uncertainty that are robust to data scarcity and outliers. We investigate two approaches. The first approach measures the variability of the output to perturbations of the model weights. The second approach captures the variability of the prediction to changes in the model architecture. We will evaluate the approaches on challenging tasks such as image completion and multilabel classification.

The connection between messaging and action is fundamental both to web applications, such as web search and sentiment analysis, and to economics. However, while prominent online applications exploit messaging in natural (human) language in order to predict non-strategic action selection, the economics literature focuses on the connection between structured stylized messaging to strategic decisions in games and multi-agent encounters. This paper aims to connect these two strands of research, which we consider highly timely and important due to the vast online textual communication on the web. Particularly, we introduce the following question: can free text expressed in natural language serve for the prediction of action selection in an economic context, modeled as a game? We initiate research on this question by providing preliminary positive results.

Landslides, movement of soil and rock under the influence of gravity, are common phenomena that cause significant human and economic losses every year. Experts use heterogeneous features such as slope, elevation, land cover, lithology, rock age, and rock family to predict landslides. To work with such features, we adapted convolutional neural networks to consider relative spatial information for the prediction task. Traditional filters in these networks either have a fixed orientation or are rotationally invariant. Intuitively, the filters should orient uphill, but there is not enough data to learn the concept of uphill; instead, it can be provided as prior knowledge. We propose a model called Locally Aligned Convolutional Neural Network, LACNN, that follows the ground surface at multiple scales to predict possible landslide occurrence for a single point. To validate our method, we created a standardized dataset of georeferenced images consisting of the heterogeneous features as inputs, and compared our method to several baselines, including linear regression, a neural network, and a convolutional network, using log-likelihood error and Receiver Operating Characteristic curves on the test set. Our model achieves 2-7% improvement in terms of accuracy and 2-15% boost in terms of log likelihood compared to the other proposed baselines.

Experience replay plays a crucial role in Reinforcement Learning (RL), enabling the agent to remember and reuse experience from the past. Most previous methods sample experience transitions using simple heuristics like uniformly sampling or prioritizing those good ones. Since humans can learn from both good and bad experiences, more sophisticated experience replay algorithms need to be developed. Inspired by the potential energy in physics, this work introduces the artificial potential field into experience replay and develops Potentialized Experience Replay (PotER) as a new and effective sampling algorithm for RL in hard exploration tasks with sparse rewards. PotER defines a potential energy function for each state in experience replay and helps the agent to learn from both good and bad experiences using intrinsic state supervision. PotER can be combined with different RL algorithms as well as the self-imitation learning algorithm. Experimental analyses and comparisons on multiple challenging hard exploration environments have verified its effectiveness and efficiency.

The positive unlabeled (PU) learning aims to train a binary classifier from a set of positive labeled samples and other unlabeled samples. Much research has been done on this special branch of weakly supervised classification problems. Since only part of the positive class is labeled, the classical PU model trains the classifier assuming the class-prior is known. However, the true class prior is usually difficult to obtain and must be learned from the given data, and the traditional methods may not work. In this paper, we formulate a convex formulation to jointly solve the class-prior unknown problem and train an accurate classifier with no need of any class-prior assumptions or additional negative samples. The class prior is estimated by pursuing the optimal solution of gradient thresholding and the classifier is simultaneously trained by performing empirical unbiased risk. The detailed derivation and theoretical analysis of the proposed model are outlined, and a comparison of our experiments with other representative methods prove the superiority of our method.

We present a new method to improve the representational power of the features in Convolutional Neural Networks (CNNs). By studying traditional image processing methods and recent CNN architectures, we propose to use positional information in CNNs for effective exploration of feature dependencies. Rather than considering feature semantics alone, we incorporate spatial positions as an augmentation for feature semantics in our design. From this vantage, we present a Position-Aware Recalibration Module (PRM in short) which recalibrates features leveraging both feature semantics and position. Furthermore, inspired by multi-head attention, our module is capable of performing multiple recalibrations where results are concatenated as the output. As PRM is efficient and easy to implement, it can be seamlessly integrated into various base networks and applied to many position-aware visual tasks. Compared to original CNNs, our PRM introduces a negligible number of parameters and FLOPs, while yielding better performance. Experimental results on ImageNet and MS COCO benchmarks show that our approach surpasses related methods by a clear margin with less computational overhead. For example, we improve the ResNet50 by absolute 1.75% (77.65% vs. 75.90%) on ImageNet 2012 validation dataset, and 1.5%~1.9% mAP on MS COCO validation dataset with almost no computational overhead. Codes are made publicly available.

Estimations on people movement behaviour within a country can provide valuable information to government strategic resource plannings. In this paper, we propose to utilize multi-domain statistical data to estimate people movements under the assumption that most population tend to move to areas with similar or better living conditions. We design a Multi-domain Matrix Factorization (MdMF) model to discover the underlying consistency patterns from these cross-domain data and estimate the movement trends using the proposed model. This research can provide important theoretical support to government and agencies in strategic resource planning and investments.

In this paper, we tackle a challenging task named video-language segmentation. Given a video and a sentence in natural language, the goal is to segment the object or actor described by the sentence in video frames. To accurately denote a target object, the given sentence usually refers to multiple attributes, such as nearby objects with spatial relations, etc. In this paper, we propose a novel Polar Relative Positional Encoding (PRPE) mechanism that represents spatial relations in a ``linguistic'' way, i.e., in terms of direction and range. Sentence feature can interact with positional embeddings in a more direct way to extract the implied relative positional relations. We also propose parameterized functions for these positional embeddings to adapt real-value directions and ranges. With PRPE, we design a Polar Attention Module (PAM) as the basic module for vision-language fusion. Our method outperforms previous best method by a large margin of 11.4% absolute improvement in terms of mAP on the challenging A2D Sentences dataset. Our method also achieves competitive performances on the J-HMDB Sentences dataset.

We investigate the problem of generating natural language summaries from knowledge base triples. Our approach is based on a pointer-generator network, which, in addition to generating regular words from a fixed target vocabulary, is able to verbalise triples in several ways. We undertake an automatic and a human evaluation on single and open-domain summaries generation tasks. Both show that our approach significantly outperforms other data-driven baselines.

Deep reinforcement learning applied to vision-based problems like Atari games maps pixels directly to actions; internally, the deep neural network bears the responsibility of both extracting useful information and making decisions based on it. By separating image processing from decision-making, one could better understand the complexity of each task, as well as potentially find smaller policy representations that are easier for humans to understand and may generalize better. To this end, we propose a new method for learning policies and compact state representations separately but simultaneously for policy approximation in reinforcement learning. State representations are generated by an encoder based on two novel algorithms: Increasing Dictionary Vector Quantization makes the encoder capable of growing its dictionary size over time, to address new observations; and Direct Residuals Sparse Coding encodes observations by aiming for highest information inclusion. We test our system on a selection of Atari games using tiny neural networks of only 6 to 18 neurons (depending on the game's controls). These are still capable of achieving results comparable---and occasionally superior---to state-of-the-art techniques which use two orders of magnitude more neurons.

In this paper we review several planning algorithms developed for zero-sum games with exponential action spaces, i.e., spaces that grow exponentially with the number of game components that can act simultaneously at a given game state. As an example, real-time strategy games have exponential action spaces because the number of actions available grows exponentially with the number of units controlled by the player. We also present a unifying perspective in which several existing algorithms can be described as an instantiation of a variant of NaiveMCTS. In addition to describing several existing planning algorithms for exponential action spaces, we show that other instantiations of this variant of NaiveMCTS represent novel and promising algorithms to be studied in future works.

Justifying a plan to a user requires answering questions about the space of possible plans. Recent work introduced a framework for doing so via plan-property dependencies, where plan properties p are Boolean functions on plans, and p entails q if all plans that satisfy p also satisfy q. We extend this work in two ways. First, we introduce new algorithms for computing plan-property dependencies, leveraging symbolic search and devising pruning methods for this purpose. Second, while the properties p were previously limited to goal facts and so-called action-set (AS) properties, here we extend them to LTL. Our new algorithms vastly outperform the previous ones, and our methods for LTL cause little overhead on AS properties.

Enumerating maximal bicliques in a bipartite graph is an important problem in data mining, with innumerable real-world applications across different domains such as web community, bioinformatics, etc. Although substantial research has been conducted on this problem, surprisingly, we find that pivot-based search space pruning, which is quite effective in clique enumeration, has not been exploited in biclique scenario. Therefore, in this paper, we explore the pivot-based pruning for biclique enumeration. We propose an algorithm for implementing the pivot-based pruning, powered by an effective index structure Containment Directed Acyclic Graph (CDAG). Meanwhile, existing literature indicates contradictory findings on the order of vertex selection in biclique enumeration. As such, we re-examine the problem and suggest an offline ordering of vertices which expedites the pivot pruning. We conduct an extensive performance study using real-world datasets from a wide range of domains. The experimental results demonstrate that our algorithm is more scalable and outperforms all the existing algorithms across all datasets and can achieve a significant speedup against the previous algorithms.

In some agent designs like inverse reinforcement learning an agent needs to learn its own reward function. Learning the reward function and optimising for it are typically two different processes, usually performed at different stages. We consider a continual (``one life'') learning approach where the agent both learns the reward function and optimises for it at the same time. We show that this comes with a number of pitfalls, such as deliberately manipulating the learning process in one direction, refusing to learn, ``learning'' facts already known to the agent, and making decisions that are strictly dominated (for all relevant reward functions). We formally introduce two desirable properties: the first is `unriggability', which prevents the agent from steering the learning process in the direction of a reward function that is easier to optimise. The second is `uninfluenceability', whereby the reward-function learning process operates by learning facts about the environment. We show that an uninfluenceable process is automatically unriggable, and if the set of possible environments is sufficiently large, the converse is true too.

In recent years, blockchain technology has created a new cryptocurrency world and has attracted a lot of attention. It also is rampant with various scams. For example, phishing scams have grabbed a lot of money and has become an important threat to users' financial security in the blockchain ecosystem. To help deal with this issue, this paper proposes a systematic approach to detect phishing accounts based on blockchain transactions and take Ethereum as an example to verify its effectiveness. Specifically, we propose a graph-based cascade feature extraction method based on transaction records and a lightGBM-based Dual-sampling Ensemble algorithm to build the identification model. Extensive experiments show that the proposed algorithm can effectively identify phishing scams.

In big cities, there are plenty of parking spaces, but we often find nowhere to park. For example, New York has 1.4 million cars and 4.4 million on-street parking spaces, but it is still not easy to find a parking place near our destination, especially during peak hours. The reason is the lack of prediction of parking behavior. If we could provide parking behavior in advance, we can ease this parking problem that affects human well-being. We observe that parking lots have periodic parking patterns, which is an important factor for parking behavior prediction. Unfortunately, existing work ignores such periodic parking patterns in parking behavior prediction, and thus incurs low accuracy. To solve this problem, we propose PewLSTM, a novel periodic weather-aware LSTM model that successfully predicts the parking behavior based on historical records, weather, environments, and weekdays. PewLSTM has been successfully integrated into a real parking space reservation system, ThsParking, which is one of the top smart parking platforms in China. Based on 452,480real parking records in 683 days from 10 parking lots, PewLSTM yields 85.3% parking prediction accuracy, which is about 20% higher than the state-of-the-art parking behavior prediction method. The code and data can be obtained fromhttps://github.com/NingxuanFeng/PewLSTM.