Multivariate time series forecasting has long received significant attention in real-world applications, such as energy consumption and traffic prediction. While recent methods demonstrate good forecasting abilities, they suffer from three fundamental limitations. (i) Discrete neural architectures: Interlacing individually parameterized spatial and temporal blocks to encode rich underlying patterns leads to discontinuous latent state trajectories and higher forecasting numerical errors. (ii) High complexity: Discrete approaches complicate models with dedicated designs and redundant parameters, leading to higher computational and memory overheads. (iii) Reliance on graph priors: Relying on predefined static graph structures limits their effectiveness and practicability in real-world applications. In this paper, we address all the above limitations by proposing a continuous model to forecast Multivariate Time series with dynamic Graph neural Ordinary Differential Equations (MTGODE). Specifically, we first abstract multivariate time series into dynamic graphs with time-evolving node features and unknown graph structures. Then, we design and solve a neural ODE to complement missing graph topologies and unify both spatial and temporal message passing, allowing deeper graph propagation and fine-grained temporal information aggregation to characterize stable and precise latent spatial-temporal dynamics. Our experiments demonstrate the superiorities of MTGODE from various perspectives on five time series benchmark datasets.
This article presents a summary graph to show the relationships between Information Retrieval (IR) and other related disciplines. The figure tells the key differences between them and the conditions under which one would transition into another.
The development of visual object tracking has continued for decades. Recent years, as the wide accessibility of the low-cost RGBD sensors, the task of visual object tracking on RGB-D videos has drawn much attention. Compared to conventional RGB-only tracking, the RGB-D videos can provide more information that facilitates objecting tracking in some complicated scenarios. The goal of this review is to summarize the relative knowledge of the research filed of RGB-D tracking. To be specific, we will generalize the related RGB-D tracking benchmarking datasets as well as the corresponding performance measurements. Besides, the existing RGB-D tracking methods are summarized in the paper. Moreover, we discuss the possible future direction in the field of RGB-D tracking.
Chinese word segmentation is necessary to provide word-level information for Chinese named entity recognition (NER) systems. However, segmentation error propagation is a challenge for Chinese NER while processing colloquial data like social media text. In this paper, we propose a model (UIcwsNN) that specializes in identifying entities from Chinese social media text, especially by leveraging ambiguous information of word segmentation. Such uncertain information contains all the potential segmentation states of a sentence that provides a channel for the model to infer deep word-level characteristics. We propose a trilogy (i.e., candidate position embedding -> position selective attention -> adaptive word convolution) to encode uncertain word segmentation information and acquire appropriate word-level representation. Experiments results on the social media corpus show that our model alleviates the segmentation error cascading trouble effectively, and achieves a significant performance improvement of more than 2% over previous state-of-the-art methods.
Molecular crystal structure prediction (CSP) seeks the most stable periodic structure given a chemical composition of a molecule and pressure-temperature conditions. Modern CSP solvers use global optimization methods to search for structures with minimal free energy within a complex energy landscape induced by intermolecular potentials. A major caveat of these methods is that initial configurations are random, making thus the search susceptible to the convergence at local minima. Providing initial configurations that are densely packed with respect to the geometric representation of a molecule can significantly accelerate CSP. Motivated by these observations we define a class of periodic packings restricted to crystallographic symmetry groups (CSG) and design a search method for densest CSG packings in an information geometric framework. Since the CSG induce a toroidal topology on the configuration space, a non-euclidean trust region method is performed on a statistical manifold consisting of probability distributions defined on an $n$-dimensional flat unit torus by extending the multivariate von Mises distribution. By introducing an adaptive quantile reformulation of the fitness function into the optimization schedule we provide the algorithm a geometric characterization through local dual geodesic flows. Moreover, we examine the geometry of the adaptive selection quantile defined trust region and show that the algorithm performs a maximization of stochastic dependence among elements of the extended multivariate von Mises distributed random vector. We experimentally evaluate its behavior and performance on various densest packings of convex polygons in $2$-dimensional CSG for which optimal solutions are known.
We present 360-DFPE, a sequential floor plan estimation method that directly takes 360-images as input without relying on active sensors or 3D information. Our approach leverages a loosely coupled integration between a monocular visual SLAM solution and a monocular 360-room layout approach, which estimate camera poses and layout geometries, respectively. Since our task is to sequentially capture the floor plan using monocular images, the entire scene structure, room instances, and room shapes are unknown. To tackle these challenges, we first handle the scale difference between visual odometry and layout geometry via formulating an entropy minimization process, which enables us to directly align 360-layouts without knowing the entire scene in advance. Second, to sequentially identify individual rooms, we propose a novel room identification algorithm that tracks every room along the camera exploration using geometry information. Lastly, to estimate the final shape of the room, we propose a shortest path algorithm with an iterative coarse-to-fine strategy, which improves prior formulations with higher accuracy and faster run-time. Moreover, we collect a new floor plan dataset with challenging large-scale scenes, providing both point clouds and sequential 360-image information. Experimental results show that our monocular solution achieves favorable performance against the current state-of-the-art algorithms that rely on active sensors and require the entire scene reconstruction data in advance. Our code and dataset will be released soon.
Configuring millimeter wave links following a conventional beam training protocol, as the one proposed in the current cellular standard, introduces a large communication overhead, specially relevant in vehicular systems, where the channels are highly dynamic. In this paper, we propose the use of a passive radar array to sense automotive radar transmissions coming from multiple vehicles on the road, and a radar processing chain that provides information about a reduced set of candidate beams for the links between the road-infrastructure and each one of the vehicles. This prior information can be later leveraged by the beam training protocol to significantly reduce overhead. The radar processing chain estimates both the timing and chirp rates of the radar signals, isolates the individual signals by filtering out interfering radar chirps, and estimates the spatial covariance of each individual radar transmission. Then, a deep network is used to translate features of these radar spatial covariances into features of the communication spatial covariances, by learning the intricate mapping between radar and communication channels, in both line-of-sight and non-line-of-sight settings. The communication rates and outage probabilities of this approach are compared against exhaustive search and pure radar-aided beam training methods (without deep learning-based mapping), and evaluated on multi-user channels simulated by ray tracing. Results show that: (i) the proposed processing chain can reliably isolate the spatial covariances for individual radars, and (ii) the radar-to-communications translation strategy based on deep learning provides a significant improvement over pure radar-aided methods in both LOS and NLOS channels.
Explaining predictions made by machine learning models is important and have attracted an increased interest. The Shapley value from cooperative game theory has been proposed as a prime approach to compute feature importances towards predictions, especially for images, text, tabular data, and recently graph neural networks (GNNs) on graphs. In this work, we revisit the appropriateness of the Shapley value for graph explanation, where the task is to identify the most important subgraph and constituent nodes for graph-level predictions. We purport that the Shapley value is a no-ideal choice for graph data because it is by definition not structure-aware. We propose a Graph Structure-aware eXplanation (GStarX) method to leverage the critical graph structure information to improve the explanation. Specifically, we propose a scoring function based on a new structure-aware value from the cooperative game theory called the HN value. When used to score node importance, the HN value utilizes graph structures to attribute cooperation surplus between neighbor nodes, resembling message passing in GNNs, so that node importance scores reflect not only the node feature importance, but also the structural roles. We demonstrate that GstarX produces qualitatively more intuitive explanations, and quantitatively improves over strong baselines on chemical graph property prediction and text graph sentiment classification.
Domain classification is the fundamental task in natural language understanding (NLU), which often requires fast accommodation to new emerging domains. This constraint makes it impossible to retrain all previous domains, even if they are accessible to the new model. Most existing continual learning approaches suffer from low accuracy and performance fluctuation, especially when the distributions of old and new data are significantly different. In fact, the key real-world problem is not the absence of old data, but the inefficiency to retrain the model with the whole old dataset. Is it potential to utilize some old data to yield high accuracy and maintain stable performance, while at the same time, without introducing extra hyperparameters? In this paper, we proposed a hyperparameter-free continual learning model for text data that can stably produce high performance under various environments. Specifically, we utilize Fisher information to select exemplars that can "record" key information of the original model. Also, a novel scheme called dynamical weight consolidation is proposed to enable hyperparameter-free learning during the retrain process. Extensive experiments demonstrate that baselines suffer from fluctuated performance and therefore useless in practice. On the contrary, our proposed model CCFI significantly and consistently outperforms the best state-of-the-art method by up to 20% in average accuracy, and each component of CCFI contributes effectively to overall performance.
We study the problem of unsupervised skill discovery, whose goal is to learn a set of diverse and useful skills with no external reward. There have been a number of skill discovery methods based on maximizing the mutual information (MI) between skills and states. However, we point out that their MI objectives usually prefer static skills to dynamic ones, which may hinder the application for downstream tasks. To address this issue, we propose Lipschitz-constrained Skill Discovery (LSD), which encourages the agent to discover more diverse, dynamic, and far-reaching skills. Another benefit of LSD is that its learned representation function can be utilized for solving goal-following downstream tasks even in a zero-shot manner - i.e., without further training or complex planning. Through experiments on various MuJoCo robotic locomotion and manipulation environments, we demonstrate that LSD outperforms previous approaches in terms of skill diversity, state space coverage, and performance on seven downstream tasks including the challenging task of following multiple goals on Humanoid. Our code and videos are available at https://shpark.me/projects/lsd/.