In this paper, we challenge the common assumption that collapsing the spatial dimensions of a 3D (spatial-channel) tensor in a convolutional neural network (CNN) into a vector via global pooling removes all spatial information. Specifically, we demonstrate that positional information is encoded based on the ordering of the channel dimensions, while semantic information is largely not. Following this demonstration, we show the real world impact of these findings by applying them to two applications. First, we propose a simple yet effective data augmentation strategy and loss function which improves the translation invariance of a CNN's output. Second, we propose a method to efficiently determine which channels in the latent representation are responsible for (i) encoding overall position information or (ii) region-specific positions. We first show that semantic segmentation has a significant reliance on the overall position channels to make predictions. We then show for the first time that it is possible to perform a `region-specific' attack, and degrade a network's performance in a particular part of the input. We believe our findings and demonstrated applications will benefit research areas concerned with understanding the characteristics of CNNs.
Despite achieving encouraging results, neural Referring Expression Generation models are often thought to lack transparency. We probed neural Referential Form Selection (RFS) models to find out to what extent the linguistic features influencing the RE form are learnt and captured by state-of-the-art RFS models. The results of 8 probing tasks show that all the defined features were learnt to some extent. The probing tasks pertaining to referential status and syntactic position exhibited the highest performance. The lowest performance was achieved by the probing models designed to predict discourse structure properties beyond the sentence level.
3D single object tracking is a key issue for robotics. In this paper, we propose a transformer module called Point-Track-Transformer (PTT) for point cloud-based 3D single object tracking. PTT module contains three blocks for feature embedding, position encoding, and self-attention feature computation. Feature embedding aims to place features closer in the embedding space if they have similar semantic information. Position encoding is used to encode coordinates of point clouds into high dimension distinguishable features. Self-attention generates refined attention features by computing attention weights. Besides, we embed the PTT module into the open-source state-of-the-art method P2B to construct PTT-Net. Experiments on the KITTI dataset reveal that our PTT-Net surpasses the state-of-the-art by a noticeable margin (~10%). Additionally, PTT-Net could achieve real-time performance (~40FPS) on NVIDIA 1080Ti GPU. Our code is open-sourced for the robotics community at https://github.com/shanjiayao/PTT.
Cross-domain object detection and semantic segmentation have witnessed impressive progress recently. Existing approaches mainly consider the domain shift resulting from external environments including the changes of background, illumination or weather, while distinct camera intrinsic parameters appear commonly in different domains, and their influence for domain adaptation has been very rarely explored. In this paper, we observe that the Field of View (FoV) gap induces noticeable instance appearance differences between the source and target domains. We further discover that the FoV gap between two domains impairs domain adaptation performance under both the FoV-increasing (source FoV < target FoV) and FoV-decreasing cases. Motivated by the observations, we propose the \textbf{Position-Invariant Transform} (PIT) to better align images in different domains. We also introduce a reverse PIT for mapping the transformed/aligned images back to the original image space and design a loss re-weighting strategy to accelerate the training process. Our method can be easily plugged into existing cross-domain detection/segmentation frameworks while bringing about negligible computational overhead. Extensive experiments demonstrate that our method can soundly boost the performance on both cross-domain object detection and segmentation for state-of-the-art techniques. Our code is available at https://github.com/sheepooo/PIT-Position-Invariant-Transform.
Robots in our daily surroundings are increasing day by day. Their usability and acceptability largely depend on their explicit and implicit interaction capability with fellow human beings. As a result, social behavior is one of the most sought-after qualities that a robot can possess. However, there is no specific aspect and/or feature that defines socially acceptable behavior and it largely depends on the situation, application, and society. In this article, we investigate one such social behavior for collocated robots. Imagine a group of people is interacting with each other and we want to join the group. We as human beings do it in a socially acceptable manner, i.e., within the group, we do position ourselves in such a way that we can participate in the group activity without disturbing/obstructing anybody. To possess such a quality, first, a robot needs to determine the formation of the group and then determine a position for itself, which we humans do implicitly. The theory of f-formation can be utilized for this purpose. As the types of formations can be very diverse, detecting the social groups is not a trivial task. In this article, we provide a comprehensive survey of the existing work on social interaction and group detection using f-formation for robotics and other applications. We also put forward a novel holistic survey framework combining all the possible concerns and modules relevant to this problem. We define taxonomies based on methods, camera views, datasets, detection capabilities and scale, evaluation approaches, and application areas. We discuss certain open challenges and limitations in current literature along with possible future research directions based on this framework. In particular, we discuss the existing methods/techniques and their relative merits and demerits, applications, and provide a set of unsolved but relevant problems in this domain.
Aspect-based sentiment analysis (ABSA), exploring sentiment polarity of aspect-given sentence, is a fine-grained task in the field of nature language processing. Previously researches typically tend to predict polarity based on the meaning of aspect and opinions. However, those approaches mainly focus on considering relations implicitly at the word level, ignore the historical impact of other positional words when the aspect appears in a certain position. Therefore, we propose a Position-based Contributive Embeddings (PosCE) to highlight the historical reference to special position aspect. Contribution of each positional words to the polarity is similar to the process of fairly distributing gains to several actors working in coalition (game theory). Therefore, we quote from the method of Shapley Value and finally gain PosCE to enhance the aspect-based representation for ABSA task. Furthermore, the PosCE can also be used for improving performances on multimodal ABSA task. Extensive experiments on both text and text-audio level using SemEval dataset show that the mainstream models advance performance in accuracy and F1 (increase 2.82% and 4.21% on average respectively) by applying our PosCE.
The RDF2vec method for creating node embeddings on knowledge graphs is based on word2vec, which, in turn, is agnostic towards the position of context words. In this paper, we argue that this might be a shortcoming when training RDF2vec, and show that using a word2vec variant which respects order yields considerable performance gains especially on tasks where entities of different classes are involved.
The identification of Alzheimer's disease (AD) and its early stages using structural magnetic resonance imaging (MRI) has been attracting the attention of researchers. Various data-driven approaches have been introduced to capture subtle and local morphological changes of the brain accompanied by the disease progression. One of the typical approaches for capturing subtle changes is patch-level feature representation. However, the predetermined regions to extract patches can limit classification performance by interrupting the exploration of potential biomarkers. In addition, the existing patch-level analyses have difficulty explaining their decision-making. To address these problems, we propose the BrainBagNet with a position-based gate (PG-BrainBagNet), a framework for jointly learning pathological region localization and AD diagnosis in an end-to-end manner. In advance, as all scans are aligned to a template in image processing, the position of brain images can be represented through the 3D Cartesian space shared by the overall MRI scans. The proposed method represents the patch-level response from whole-brain MRI scans and discriminative brain-region from position information. Based on the outcomes, the patch-level class evidence is calculated, and then the image-level prediction is inferred by a transparent aggregation. The proposed models were evaluated on the ADNI datasets. In five-fold cross-validation, the classification performance of the proposed method outperformed that of the state-of-the-art methods in both AD diagnosis (AD vs. normal control) and mild cognitive impairment (MCI) conversion prediction (progressive MCI vs. stable MCI) tasks. In addition, changes in the identified discriminant regions and patch-level class evidence according to the patch size used for model training are presented and analyzed.
GNNs have been proven to perform highly effective in various node-level, edge-level, and graph-level prediction tasks in several domains. Existing approaches mainly focus on static graphs. However, many graphs change over time with their edge may disappear, or node/edge attribute may alter from one time to the other. It is essential to consider such evolution in representation learning of nodes in time varying graphs. In this paper, we propose a Temporal Multi-layered Position-aware Graph Neural Network (TMP-GNN), a node embedding approach for dynamic graph that incorporates the interdependence of temporal relations into embedding computation. We evaluate the performance of TMP-GNN on two different representations of temporal multilayered graphs. The performance is assessed against the most popular GNNs on node-level prediction tasks. Then, we incorporate TMP-GNN into a deep learning framework to estimate missing data and compare the performance with their corresponding competent GNNs from our former experiment, and a baseline method. Experimental results on four real-world datasets yield up to 58% of lower ROC AUC for pairwise node classification task, and 96% of lower MAE in missing feature estimation, particularly for graphs with a relatively high number of nodes and lower mean degree of connectivity.
This paper introduces Summary Explorer, a new tool to support the manual inspection of text summarization systems by compiling the outputs of 55~state-of-the-art single document summarization approaches on three benchmark datasets, and visually exploring them during a qualitative assessment. The underlying design of the tool considers three well-known summary quality criteria (coverage, faithfulness, and position bias), encapsulated in a guided assessment based on tailored visualizations. The tool complements existing approaches for locally debugging summarization models and improves upon them. The tool is available at https://tldr.webis.de/
Future cellular networks that utilize millimeter wave signals provide new opportunities in positioning and situational awareness. Large bandwidths combined with large antenna arrays provide unparalleled delay and angle resolution, allowing high accuracy localization but also building up a map of the environment. Even the most basic filter intended for simultaneous localization and mapping exhibits high computational overhead since the methods rely on sigma point or particle-based approximations. In this paper, a first order Taylor series based Gaussian approximation of the filtering distribution is used and it is demonstrated that the developed extended Kalman probability hypothesis density filter is computationally very efficient. In addition, the results imply that efficiency does not come with the expense of estimation accuracy since the method nearly achieves the position error bound.
We observe a severe under-reporting of the different kinds of errors that Natural Language Generation systems make. This is a problem, because mistakes are an important indicator of where systems should still be improved. If authors only report overall performance metrics, the research community is left in the dark about the specific weaknesses that are exhibited by `state-of-the-art' research. Next to quantifying the extent of error under-reporting, this position paper provides recommendations for error identification, analysis and reporting.
For robots to operate in a three dimensional world and interact with humans, learning spatial relationships among objects in the surrounding is necessary. Reasoning about the state of the world requires inputs from many different sensory modalities including vision ($V$) and haptics ($H$). We examine the problem of desk organization: learning how humans spatially position different objects on a planar surface according to organizational ''preference''. We model this problem by examining how humans position objects given multiple features received from vision and haptic modalities. However, organizational habits vary greatly between people both in structure and adherence. To deal with user organizational preferences, we add an additional modality, ''utility'' ($U$), which informs on a particular human's perceived usefulness of a given object. Models were trained as generalized (over many different people) or tailored (per person). We use two types of models: random forests, which focus on precise multi-task classification, and Markov logic networks, which provide an easily interpretable insight into organizational habits. The models were applied to both synthetic data, which proved to be learnable when using fixed organizational constraints, and human-study data, on which the random forest achieved over 90% accuracy. Over all combinations of $\{H, U, V\}$ modalities, $UV$ and $HUV$ were the most informative for organization. In a follow-up study, we gauged participants preference of desk organizations by a generalized random forest organization vs. by a random model. On average, participants rated the random forest models as 4.15 on a 5-point Likert scale compared to 1.84 for the random model
Transformers have made great progress in dealing with computer vision tasks. However, existing vision transformers do not yet possess the ability of building the interactions among features of different scales, which is perceptually important to visual inputs. The reasons are two-fold: (1) Input embeddings of each layer are equal-scale, so no cross-scale feature can be extracted; (2) to lower the computational cost, some vision transformers merge adjacent embeddings inside the self-attention module, thus sacrificing small-scale (fine-grained) features of the embeddings and also disabling the cross-scale interactions. To this end, we propose Cross-scale Embedding Layer (CEL) and Long Short Distance Attention (LSDA). On the one hand, CEL blends each embedding with multiple patches of different scales, providing the self-attention module itself with cross-scale features. On the other hand, LSDA splits the self-attention module into a short-distance one and a long-distance counterpart, which not only reduces the computational burden but also keeps both small-scale and large-scale features in the embeddings. Through the above two designs, we achieve cross-scale attention. Besides, we put forward a dynamic position bias for vision transformers to make the popular relative position bias apply to variable-sized images. Hinging on the cross-scale attention module, we construct a versatile vision architecture, dubbed CrossFormer, which accommodates variable-sized inputs. Extensive experiments show that CrossFormer outperforms the other vision transformers on image classification, object detection, instance segmentation, and semantic segmentation tasks. The code has been released: https://github.com/cheerss/CrossFormer.
Weakly-Sticky(WS) Datalog+/- is an expressive member of the family of Datalog+/- program classes that is defined on the basis of the conditions of stickiness and weak-acyclicity. Conjunctive query answering (QA) over the WS programs has been investigated, and its tractability in data complexity has been established. However, the design and implementation of practical QA algorithms and their optimizations have been open. In order to fill this gap, we first study Sticky and WS programs from the point of view of the behavior of the chase procedure. We extend the stickiness property of the chase to that of generalized stickiness of the chase (GSCh) modulo an oracle that selects (and provides) the predicate positions where finitely values appear during the chase. Stickiness modulo a selection function S that provides only a subset of those positions defines sch(S), a semantic subclass of GSCh. Program classes with selection functions include Sticky and WS, and another syntactic class that we introduce and characterize, namely JWS, of jointly-weakly-sticky programs, which contains WS. The selection functions for these last three classes are computable, and no external, possibly non-computable oracle is needed. We propose a bottom-up QA algorithm for programs in the class sch(S), for a general selection function S. As a particular case, we obtain a polynomial-time QA algorithm for JWS and weakly-sticky programs. Unlike WS, JWS turns out to be closed under magic-sets query optimization. As a consequence, both the generic polynomial-time QA algorithm and its magic-set optimization can be particularized and applied to WS.
Omni-directional mobile robot (OMR) systems have been very popular in academia and industry for their superb maneuverability and flexibility. Yet their potential has not been fully exploited, where the extra degree of freedom in OMR can potentially enable the robot to carry out extra tasks. For instance, gimbals or sensors on robots may suffer from a limited field of view or be constrained by the inherent mechanical design, which will require the chassis to be orientation-aware and respond in time. To solve this problem and further develop the OMR systems, in this paper, we categorize the tasks related to OMR chassis into orientation transition tasks and position transition tasks, where the two tasks can be carried out at the same time. By integrating the parallel task goals in a single planning problem, we proposed an orientation-aware planning architecture for OMR systems to execute the orientation transition and position transition in a unified and efficient way. A modified trajectory optimization method called orientation-aware timed-elastic-band (OATEB) is introduced to generate the trajectory that satisfies the requirements of both tasks. Experiments in both 2D simulated environments and real scenes are carried out. A four-wheeled OMR is deployed to conduct the real scene experiment and the results demonstrate that the proposed method is capable of simultaneously executing parallel tasks and is applicable to real-life scenarios.
Chinese sentiment analysis (CSA) has always been one of the challenges in natural language processing due to its complexity and uncertainty. Transformer has succeeded in capturing semantic features, but it uses position encoding to capture sequence features, which has great shortcomings compared with the recurrent model. In this paper, we propose T-E-GRU for Chinese sentiment analysis, which combine transformer encoder and GRU. We conducted experiments on three Chinese comment datasets. In view of the confusion of punctuation marks in Chinese comment texts, we selectively retain some punctuation marks with sentence segmentation ability. The experimental results show that T-E-GRU outperforms classic recurrent model and recurrent model with attention.
Currently there is no standard indoor positioning system, similar to outdoor GPS. However, WiFi signals have been used in a large number of proposals to achieve the above positioning, many of which use machine learning to do so. But what are the most commonly used techniques in machine learning? What accuracy do they achieve? Where have they been tested? This article presents a systematic review of works between 2019 and 2021 that use WiFi as the signal for positioning and machine learning models to estimate indoor position. 64 papers have been identified as relevant, which have been systematically analyzed for a better understanding of the current situation in different aspects. The results show that indoor positioning based on WiFi trends use neural network-based models, evaluated in empirical experiments. Despite this, many works still conduct an assessment in small areas, which can influence the goodness of the results presented.
The paper describes a method for measuring the similarity and symmetry of an image annotated with bounding boxes indicating image objects. The latter representation became popular recently due to the rapid development of fast and efficient deep-learning-based object-detection methods. The proposed approach allows for comparing sets of bounding boxes to estimate the degree of similarity of their underlying images. It is based on the fuzzy approach that uses the fuzzy mutual position (FMP) matrix to describe spatial composition and relations between bounding boxes within an image. A method of computing the similarity of two images described by their FMP matrices is proposed and the algorithm of its computation. It outputs the single scalar value describing the degree of content-based image similarity. By modifying the method`s parameters, instead of similarity, the reflectional symmetry of object composition may also be measured. The proposed approach allows for measuring differences in objects` composition of various intensities. It is also invariant to translation and scaling and - in case of symmetry detection - position and orientation of the symmetry axis. A couple of examples illustrate the method.
This paper presents a novel dataset for the development of visual navigation and simultaneous localisation and mapping (SLAM) algorithms as well as for underwater intervention tasks. It differs from existing datasets as it contains ground truth for the vehicle's position captured by an underwater motion tracking system. The dataset contains distortion-free and rectified stereo images along with the calibration parameters of the stereo camera setup. Furthermore, the experiments were performed and recorded in a controlled environment, where current and waves could be generated allowing the dataset to cover a wide range of conditions - from calm water to waves and currents of significant strength.