IJCAI 2020 Papers

Recent advances cast the entity-relation extraction to a multi-turn question answering (QA) task and provide an effective solution based on the machine reading comprehension (MRC) models. However, they use a single question to characterize the meaning of entities and relations, which is intuitively not enough because of the variety of context semantics. Meanwhile, existing models enumerate all relation types to generate questions, which is inefficient and easily leads to confusing questions. In this paper, we improve the existing MRC-based entity-relation extraction model through diverse question answering. First, a diversity question answering mechanism is introduced to detect entity spans and two answering selection strategies are designed to integrate different answers. Then, we propose to predict a subset of potential relations and filter out irrelevant ones to generate questions effectively. Finally, entity and relation extractions are integrated in an end-to-end way and optimized through joint learning. Experiment results show that the proposed method significantly outperforms baseline models, which improves the relation F1 to 62.1% (+1.9%) on ACE05 and 71.9% (+3.0%) on CoNLL04. Our implementation is available at https://github.com/TanyaZhao/MRC4ERE.

Natural language processing (NLP) models are known vulnerable to adversarial examples, similar to image processing models. Studying adversarial texts is an essential step to improve the robustness of NLP models. However, existing studies mainly focus on analyzing English texts and generating adversarial examples for English texts. There is no work studying the possibility and effect of the transformation to another language, e.g, Chinese. In this paper, we analyze the differences between Chinese and English, and explore the methodology to transform the existing English adversarial generation method to Chinese. We propose a novel black-box adversarial Chinese texts generation solution Argot, by utilizing the method for adversarial English samples and several novel methods developed on Chinese characteristics. Argot could effectively and efficiently generate adversarial Chinese texts with good readability. Furthermore, Argot could also automatically generate targeted Chinese adversarial text, achieving a high success rate and ensuring readability of the Chinese.

Talking face generation aims to synthesize a face video with precise lip synchronization as well as a smooth transition of facial motion over the entire video via the given speech clip and facial image. Most existing methods mainly focus on either disentangling the information in a single image or learning temporal information between frames. However, cross-modality coherence between audio and video information has not been well addressed during synthesis. In this paper, we propose a novel arbitrary talking face generation framework by discovering the audio-visual coherence via the proposed Asymmetric Mutual Information Estimator (AMIE). In addition, we propose a Dynamic Attention (DA) block by selectively focusing the lip area of the input image during the training stage, to further enhance lip synchronization. Experimental results on benchmark LRW dataset and GRID dataset transcend the state-of-the-art methods on prevalent metrics with robust high-resolution synthesizing on gender and pose variations.

We study the symmetric weighted first-order model counting task and present ApproxWFOMC, a novel anytime method for efficiently bounding the weighted first-order model count of a sentence given an unweighted first-order model counting oracle. The algorithm has applications to inference in a variety of first-order probabilistic representations, such as Markov logic networks and probabilistic logic programs. Crucially for many applications, no assumptions are made on the form of the input sentence. Instead, the algorithm makes use of the symmetry inherent in the problem by imposing cardinality constraints on the number of possible true groundings of a sentence's literals. Realising the first-order model counting oracle in practice using the approximate hashing-based model counter ApproxMC3, we show how our algorithm is competitive with existing approximate and exact techniques for inference in first-order probabilistic models. We additionally provide PAC guarantees on the accuracy of the bounds generated.

In fair division of indivisible goods, finding an allocation that satisfies fairness and efficiency simultaneously is highly desired but computationally hard. We solve this problem approximately in polynomial time by modeling it as a bi-criteria optimization problem that can be solved efficiently by determining an approximate Pareto set of bounded size. We focus on two criteria: max-min fairness and utilitarian efficiency, and study this problem for the setting when there are only a few item types or a few agent types. We show in both cases that one can construct an approximate Pareto set in time polynomial in the input size, either by designing a dynamic programming scheme, or a linear-programming algorithm. Our techniques strengthen known methods and can be potentially applied to other notions of fairness and efficiency as well.

Ontology-mediated query answering (OMQA) is a promising approach to data access and integration that has been actively studied in the knowledge representation and database communities for more than a decade. The vast majority of work on OMQA focuses on conjunctive queries, whereas more expressive queries that feature counting or other forms of aggregation remain largely unexplored. In this paper, we introduce a general form of counting query, relate it to previous proposals, and study the complexity of answering such queries in the presence of DL-Lite ontologies. As it follows from existing work that query answering is intractable and often of high complexity, we consider some practically relevant restrictions, for which we establish improved complexity bounds.

Machine reading comprehension methods that gen- erate answers by referring to multiple passages for a question have gained much attention in AI and NLP communities. The current methods, however, do not investigate the relationships among multi- ple passages in the answer generation process, even though topics correlated among the passages may be answer candidates. Our method, called neural answer Generation through Unified Memories over Multiple Passages (GUM-MP), solves this problem as follows. First, it determines which tokens in the passages are matched to the question. In particular, it investigates matches between tokens in positive passages, which are assigned to the question, and those in negative passages, which are not related to the question. Next, it determines which tokens in the passage are matched to other passages assigned to the same question and at the same time it investi- gates the topics in which they are matched. Finally, it encodes the token sequences with the above two matching results into unified memories in the pas- sage encoders and learns the answer sequence by using an encoder-decoder with a multiple-pointer- generator mechanism. As a result, GUM-MP can generate answers by pointing to important tokens present across passages. Evaluations indicate that GUM-MP generates much more accurate results than the current models do.

Existing convergence analyses of Q-learning mostly focus on the vanilla stochastic gradient descent (SGD) type of updates. Despite the Adaptive Moment Estimation (Adam) has been commonly used for practical Q-learning algorithms, there has not been any convergence guarantee provided for Q-learning with such type of updates. In this paper, we first characterize the convergence rate for Q-AMSGrad, which is the Q-learning algorithm with AMSGrad update (a commonly adopted alternative of Adam for theoretical analysis). To further improve the performance, we propose to incorporate the momentum restart scheme to Q-AMSGrad, resulting in the so-called Q-AMSGradR algorithm. The convergence rate of Q-AMSGradR is also established. Our experiments on a linear quadratic regulator problem demonstrate that the two proposed Q-learning algorithms outperform the vanilla Q-learning with SGD updates. The two algorithms also exhibit significantly better performance than the DQN learning method over a batch of Atari 2600 games.

Analogical proportions are statements of the form “x is to y as z is to t”, where x, y, z, t are items of the same nature, or not. In this paper, we more particularly consider “relational proportions” of the form “object A has the same relationship with attribute a as object B with attribute b”. We provide a formal definition for relational proportions, and investigate how they can be extracted from a formal context, in the setting of formal concept analysis.

Energy-efficient navigation constitutes an important challenge in electric vehicles, due to their limited battery capacity. We employ a Bayesian approach to model the energy consumption at road segments for efficient navigation. In order to learn the model parameters, we develop an online learning framework and investigate several exploration strategies such as Thompson Sampling and Upper Confidence Bound. We then extend our online learning framework to multi-agent setting, where multiple vehicles adaptively navigate and learn the parameters of the energy model. We analyze Thompson Sampling and establish rigorous regret bounds on its performance. Finally, we demonstrate the performance of our methods via several real-world experiments on Luxembourg SUMO Traffic dataset.

The lack of sufficient training data in many domains, poses a major challenge to the construction of domain-specific machine reading comprehension (MRC) models with satisfying performance. In this paper, we propose a novel iterative multi-source mutual knowledge transfer framework for MRC. As an extension of the conventional knowledge transfer with one-to-one correspondence, our framework focuses on the many-to-many mutual transfer, which involves synchronous executions of multiple many-to-one transfers in an iterative manner.Specifically, to update a target-domain MRC model, we first consider other domain-specific MRC models as individual teachers, and employ knowledge distillation to train a multi-domain MRC model, which is differentially required to fit the training data and match the outputs of these individual models according to their domain-level similarities to the target domain. After being initialized by the multi-domain MRC model, the target-domain MRC model is fine-tuned to match both its training data and the output of its previous best model simultaneously via knowledge distillation. Compared with previous approaches, our framework can continuously enhance all domain-specific MRC models by enabling each model to iteratively and differentially absorb the domain-shared knowledge from others. Experimental results and in-depth analyses on several benchmark datasets demonstrate the effectiveness of our framework.

Discovering tractable mathematical expressions that best explain a dataset is a long-standing challenge in artificial intelligence. This problem, known as symbolic regression, is relevant when one seeks to generate new physical knowledge and insights. Since practitioners are primarily interested in knowledge generation, the ability to interact with a symbolic regression algorithm would be highly valuable. Thus, we present an interactive symbolic regression framework that allows users not only to configure runs, but also to control the system during training. The interface provides real-time visualization and diagnostics to help guide the user as they control the algorithm on the fly.

Studies on next point-of-interest (POI) recommendation mainly seek to learn users' transition patterns with certain historical check-ins. However, in reality, users' movements are typically uncertain (i.e., fuzzy and incomplete) where most existing methods suffer from the transition pattern vanishing issue. To ease this issue, we propose a novel interactive multi-task learning (iMTL) framework to better exploit the interplay between activity and location preference. Specifically, iMTL introduces: (1) temporal-aware activity encoder equipped with fuzzy characterization over uncertain check-ins to unveil the latent activity transition patterns; (2) spatial-aware location preference encoder to capture the latent location transition patterns; and (3) task-specific decoder to make use of the learned latent transition patterns and enhance both activity and location prediction tasks in an interactive manner. Extensive experiments on three real-world datasets show the superiority of iMTL.

There is growing interest in low rank representation (LRR) for subspace clustering. Existing latent LRR methods can exploit the global structure of data when the observations are insufficient and/or grossly corrupted, but it cannot capture the intrinsic structure due to the neglect of the local information of data. In this paper, we proposed an improved latent LRR model with a distance regularization and a non-negative regularization jointly, which can effectively discover the global and local structure of data for graph learning and improve the expression of the model. Then, an efficiently iterative algorithm is developed to optimize the improved latent LRR model. In addition, traditional subspace clustering characterizes a fixed numbers of cluster, which cannot efficiently make model selection. An efficiently automatic subspace clustering is developed via the bias and variance trade-off, where the numbers of cluster can be automatically added and discarded on the fly.

The trend of moving online in the retail industry has created great pressure for the logistics industry to catch up both in terms of volume and response time. On one hand, volume is fluctuating at greater magnitude, making peaks higher; on the other hand, customers are also expecting shorter response time. As a result, logistics service providers are pressured to expand and keep up with the demands. Expanding fleet capacity, however, is not sustainable as capacity built for the peak seasons would be mostly vacant during ordinary days. One promising solution is to engage crowdsourced workers, who are not employed full-time but would be willing to help with the deliveries if their schedules permit. The challenge, however, is to choose appropriate sets of tasks that would not cause too much disruption from their intended routes, while satisfying each delivery task's delivery time window requirement. In this paper, we propose a decision-support algorithm to select delivery tasks for a single crowdsourced worker that best fit his/her upcoming route both in terms of additional travel time and the time window requirements at all stops along his/her route, while at the same time satisfies tasks' delivery time windows. Our major contributions are in the formulation of the problem and the design of an efficient exact algorithm based on the branch-and-cut approach. The major innovation we introduce is the efficient generation of promising valid inequalities via our separation heuristics. In all numerical instances we study, our approach manages to reach optimality yet with much fewer computational resource requirement than the plain integer linear programming formulation. The greedy heuristic, while efficient in time, only achieves around 40-60% of the optimum in all cases. To illustrate how our solver could help in advancing the sustainability objective, we also quantify the reduction in the carbon footprint.

In this article, we propose an end-to-end adaptive framework for optimal trade execution based on Proximal Policy Optimization (PPO). We use two methods to account for the time dependencies in the market data based on two different neural network architecture: 1) Long short-term memory (LSTM) networks, 2) Fully-connected networks (FCN) by stacking the most recent limit orderbook (LOB) information as model inputs. The proposed framework can make trade execution decisions based on level-2 limit order book (LOB) information such as bid/ask prices and volumes directly without manually designed attributes as in previous research. Furthermore, we use a sparse reward function, which gives the agent reward signals at the end of each episode as an indicator of its relative performances against the baseline model, rather than implementation shortfall (IS) or a shaped reward function. The experimental results have demonstrated advantages over IS and the shaped reward function in terms of performance and simplicity. The proposed framework has outperformed the industry commonly used baseline models such as TWAP, VWAP, and AC as well as several Deep Reinforcement Learning (DRL) models on most of the 14 US equities in our experiments.

Conversion rate (CVR) prediction is becoming increasingly important in the multi-billion dollar online display advertising industry. It has two major challenges: firstly, the scarce user history data is very complicated and non-linear; secondly, the time delay between the clicks and the corresponding conversions can be very large, e.g., ranging from seconds to weeks. Existing models usually suffer from such scarce and delayed conversion behaviors. In this paper, we propose a novel deep learning framework to tackle the two challenges. Specifically, we extract the pre-trained embedding from impressions/clicks to assist in conversion models and propose an inner/self-attention mechanism to capture the fine-grained personalized product purchase interests from the sequential click data. Besides, to overcome the time-delay issue, we calibrate the delay model by learning dynamic hazard function with the abundant post-click data more in line with the real distribution. Empirical experiments with real-world user behavior data prove the effectiveness of the proposed method.

Anomaly detection in data storage systems is a challenging problem due to the high dimensional sequential data involved, and lack of labels. The state of the art for automating anomaly detection in these systems typically relies on hand crafted rules and thresholds which mainly allow to distinguish between normal and abnormal behavior of each indicator in isolation. In this work we present an end-to-end framework based on convolutional autoencoders which not only allows for anomaly detection on multivariate time series data, but also provides explainability. This is done by identifying similar historic anomalies and extracting the most influential indicators. These are then presented to relevant personnel such as system designers and architects, or to support engineers for further analysis. We demonstrate the application of this framework along with an intuitive interactive web interface which was developed for data storage system anomaly detection. We discuss how this framework along with its explainability aspects enables support engineers to effectively tackle abnormal behaviors, all while allowing for crucial feedback.

Bipartite b-matching, where agents on one side of a market are matched to one or more agents or items on the other, is a classical model that is used in myriad application areas such as healthcare, advertising, education, and general resource allocation. Traditionally, the primary goal of such models is to maximize a linear function of the constituent matches (e.g., linear social welfare maximization) subject to some constraints. Recent work has studied a new goal of balancing whole-match diversity and economic efficiency, where the objective is instead a monotone submodular function over the matching. Basic versions of this problem are solvable in polynomial time. In this work, we prove that the problem of simultaneously maximizing diversity along several features (e.g., country of citizenship, gender, skills) is NP-hard. To address this problem, we develop the first combinatorial algorithm that constructs provably-optimal diverse b-matchings in pseudo-polynomial time. We also provide a Mixed-Integer Quadratic formulation for the same problem and show that our method guarantees optimal solutions and takes less computation time for a reviewer assignment application. The source code is made available at https://github.com/faezahmed/diverse_matching.