PyTorchKR Papers

최신 논문

총 715편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2109.01238공식 코드 · Papers with Code

An Empirical Study on Leveraging Position Embeddings for Target-oriented Opinion Words Extraction

Target-oriented opinion words extraction (TOWE) (Fan et al., 2019b) is a new subtask of target-oriented sentiment analysis that aims to extract opinion words for a given aspect in text. Current state-of-the-art methods leverage position embeddings to capture the relative position of a word to the target. However, the performance of these methods depends on the ability to incorporate this information into word representations. In this paper, we explore a variety of text encoders based on pretrained word embeddings or language models that leverage part-of-speech and position embeddings, aiming to examine the actual contribution of each component in TOWE. We also adapt a graph convolutional network (GCN) to enhance word representations by incorporating syntactic information. Our experimental results demonstrate that BiLSTM-based models can effectively encode position information into word representations while using a GCN only achieves marginal gains. Interestingly, our simple methods outperform several state-of-the-art complex neural structures.

arXiv:2108.13202공식 코드 · Papers with Code

PTRAIL -- A python package for parallel trajectory data preprocessing

Trajectory data represent a trace of an object that changes its position in space over time. This kind of data is complex to handle and analyze, since it is generally produced in huge quantities, often prone to errors generated by the geolocation device, human mishandling, or area coverage limitation. Therefore, there is a need for software specifically tailored to preprocess trajectory data. In this work we propose PTRAIL, a python package offering several trajectory preprocessing steps, including filtering, feature extraction, and interpolation. PTRAIL uses parallel computation and vectorization, being suitable for large datasets and fast compared to other python libraries.

arXiv:2108.13344공식 코드 · Papers with Code

Enlisting 3D Crop Models and GANs for More Data Efficient and Generalizable Fruit Detection

Training real-world neural network models to achieve high performance and generalizability typically requires a substantial amount of labeled data, spanning a broad range of variation. This data-labeling process can be both labor and cost intensive. To achieve desirable predictive performance, a trained model is typically applied into a domain where the data distribution is similar to the training dataset. However, for many agricultural machine learning problems, training datasets are collected at a specific location, during a specific period in time of the growing season. Since agricultural systems exhibit substantial variability in terms of crop type, cultivar, management, seasonal growth dynamics, lighting condition, sensor type, etc, a model trained from one dataset often does not generalize well across domains. To enable more data efficient and generalizable neural network models in agriculture, we propose a method that generates photorealistic agricultural images from a synthetic 3D crop model domain into real world crop domains. The method uses a semantically constrained GAN (generative adversarial network) to preserve the fruit position and geometry. We observe that a baseline CycleGAN method generates visually realistic target domain images but does not preserve fruit position information while our method maintains fruit positions well. Image generation results in vineyard grape day and night images show the visual outputs of our network are much better compared to a baseline network. Incremental training experiments in vineyard grape detection tasks show that the images generated from our method can significantly speed the domain adaption process, increase performance for a given number of labeled images (i.e. data efficiency), and decrease labeling requirements.

arXiv:2108.12409공식 코드 · Papers with Code

Train Short, Test Long: Attention with Linear Biases Enables Input Length Extrapolation

Since the introduction of the transformer model by Vaswani et al. (2017), a fundamental question has yet to be answered: how does a model achieve extrapolation at inference time for sequences that are longer than it saw during training? We first show that extrapolation can be enabled by simply changing the position representation method, though we find that current methods do not allow for efficient extrapolation. We therefore introduce a simpler and more efficient position method, Attention with Linear Biases (ALiBi). ALiBi does not add positional embeddings to word embeddings; instead, it biases query-key attention scores with a penalty that is proportional to their distance. We show that this method trains a 1.3 billion parameter model on input sequences of length 1024 that extrapolates to input sequences of length 2048, achieving the same perplexity as a sinusoidal position embedding model trained on inputs of length 2048 but training 11% faster and using 11% less memory. ALiBi's inductive bias towards recency also leads it to outperform multiple strong position methods on the WikiText-103 benchmark.

arXiv:2108.09779공식 코드 · Papers with Code

Transferring Dexterous Manipulation from GPU Simulation to a Remote Real-World TriFinger

We present a system for learning a challenging dexterous manipulation task involving moving a cube to an arbitrary 6-DoF pose with only 3-fingers trained with NVIDIA's IsaacGym simulator. We show empirical benefits, both in simulation and sim-to-real transfer, of using keypoints as opposed to position+quaternion representations for the object pose in 6-DoF for policy observations and in reward calculation to train a model-free reinforcement learning agent. By utilizing domain randomization strategies along with the keypoint representation of the pose of the manipulated object, we achieve a high success rate of 83% on a remote TriFinger system maintained by the organizers of the Real Robot Challenge. With the aim of assisting further research in learning in-hand manipulation, we make the codebase of our system, along with trained checkpoints that come with billions of steps of experience available, at https://s2r2-ig.github.io

arXiv:2108.08538공식 코드 · Papers with Code

Mixture-Based Correction for Position and Trust Bias in Counterfactual Learning to Rank

In counterfactual learning to rank (CLTR) user interactions are used as a source of supervision. Since user interactions come with bias, an important focus of research in this field lies in developing methods to correct for the bias of interactions. Inverse propensity scoring (IPS) is a popular method suitable for correcting position bias. Affine correction (AC) is a generalization of IPS that corrects for position bias and trust bias. IPS and AC provably remove bias, conditioned on an accurate estimation of the bias parameters. Estimating the bias parameters, in turn, requires an accurate estimation of the relevance probabilities. This cyclic dependency introduces practical limitations in terms of sensitivity, convergence and efficiency. We propose a new correction method for position and trust bias in CLTR in which, unlike the existing methods, the correction does not rely on relevance estimation. Our proposed method, mixture-based correction (MBC), is based on the assumption that the distribution of the CTRs over the items being ranked is a mixture of two distributions: the distribution of CTRs for relevant items and the distribution of CTRs for non-relevant items. We prove that our method is unbiased. The validity of our proof is not conditioned on accurate bias parameter estimation. Our experiments show that MBC, when used in different bias settings and accompanied by different LTR algorithms, outperforms AC, the state-of-the-art method for correcting position and trust bias, in some settings, while performing on par in other settings. Furthermore, MBC is orders of magnitude more efficient than AC in terms of the training time.

arXiv:2108.07884공식 코드 · Papers with Code

Global Pooling, More than Meets the Eye: Position Information is Encoded Channel-Wise in CNNs

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.

arXiv:2108.06455공식 코드 · Papers with Code

PTT: Point-Track-Transformer Module for 3D Single Object Tracking in Point Clouds

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.

arXiv:2108.07142공식 코드 · Papers with Code

PIT: Position-Invariant Transform for Cross-FoV Domain Adaptation

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.

arXiv:2108.05280공식 코드 · Papers with Code

Putting RDF2vec in Order

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.

arXiv:2108.04555공식 코드 · Papers with Code

Deep Joint Learning of Pathological Region Localization and Alzheimer's Disease Diagnosis

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.

arXiv:2108.01879공식 코드 · Papers with Code

Summary Explorer: Visualizing the State of the Art in Text Summarization

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/

arXiv:2108.00154공식 코드 · Papers with Code

CrossFormer: A Versatile Vision Transformer Hinging on Cross-scale Attention

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.

arXiv:2107.14222공식 코드 · Papers with Code

Rethinking and Improving Relative Position Encoding for Vision Transformer

Relative position encoding (RPE) is important for transformer to capture sequence ordering of input tokens. General efficacy has been proven in natural language processing. However, in computer vision, its efficacy is not well studied and even remains controversial, e.g., whether relative position encoding can work equally well as absolute position? In order to clarify this, we first review existing relative position encoding methods and analyze their pros and cons when applied in vision transformers. We then propose new relative position encoding methods dedicated to 2D images, called image RPE (iRPE). Our methods consider directional relative distance modeling as well as the interactions between queries and relative position embeddings in self-attention mechanism. The proposed iRPE methods are simple and lightweight. They can be easily plugged into transformer blocks. Experiments demonstrate that solely due to the proposed encoding methods, DeiT and DETR obtain up to 1.5% (top-1 Acc) and 1.3% (mAP) stable improvements over their original versions on ImageNet and COCO respectively, without tuning any extra hyperparameters such as learning rate and weight decay. Our ablation and analysis also yield interesting findings, some of which run counter to previous understanding. Code and models are open-sourced at https://github.com/microsoft/Cream/tree/main/iRPE.

arXiv:2107.08653공식 코드 · Papers with Code

Cell Detection in Domain Shift Problem Using Pseudo-Cell-Position Heatmap

The domain shift problem is an important issue in automatic cell detection. A detection network trained with training data under a specific condition (source domain) may not work well in data under other conditions (target domain). We propose an unsupervised domain adaptation method for cell detection using the pseudo-cell-position heatmap, where a cell centroid becomes a peak with a Gaussian distribution in the map. In the prediction result for the target domain, even if a peak location is correct, the signal distribution around the peak often has anon-Gaussian shape. The pseudo-cell-position heatmap is re-generated using the peak positions in the predicted heatmap to have a clear Gaussian shape. Our method selects confident pseudo-cell-position heatmaps using a Bayesian network and adds them to the training data in the next iteration. The method can incrementally extend the domain from the source domain to the target domain in a semi-supervised manner. In the experiments using 8 combinations of domains, the proposed method outperformed the existing domain adaptation methods.

arXiv:2107.05566공식 코드 · Papers with Code

ProGS: Property Graph Shapes Language (Extended Version)

Property graphs constitute data models for representing knowledge graphs. They allow for the convenient representation of facts, including facts about facts, represented by triples in subject or object position of other triples. Knowledge graphs such as Wikidata are created by a diversity of contributors and a range of sources leaving them prone to two types of errors. The first type of error, falsity of facts, is addressed by property graphs through the representation of provenance and validity, making triples occur as first-order objects in subject position of metadata triples. The second type of error, violation of domain constraints, has not been addressed with regard to property graphs so far. In RDF representations, this error can be addressed by shape languages such as SHACL or ShEx, which allow for checking whether graphs are valid with respect to a set of domain constraints. Borrowing ideas from the syntax and semantics definitions of SHACL, we design a shape language for property graphs, ProGS, which allows for formulating shape constraints on property graphs including their specific constructs, such as edges with identities and key-value annotations to both nodes and edges. We define a formal semantics of ProGS, investigate the resulting complexity of validating property graphs against sets of ProGS shapes, compare with corresponding results for SHACL, and implement a prototypical validator that utilizes answer set programming.

arXiv:2107.05235공식 코드 · Papers with Code

Position-enhanced and Time-aware Graph Convolutional Network for Sequential Recommendations

Most of the existing deep learning-based sequential recommendation approaches utilize the recurrent neural network architecture or self-attention to model the sequential patterns and temporal influence among a user's historical behavior and learn the user's preference at a specific time. However, these methods have two main drawbacks. First, they focus on modeling users' dynamic states from a user-centric perspective and always neglect the dynamics of items over time. Second, most of them deal with only the first-order user-item interactions and do not consider the high-order connectivity between users and items, which has recently been proved helpful for the sequential recommendation. To address the above problems, in this article, we attempt to model user-item interactions by a bipartite graph structure and propose a new recommendation approach based on a Position-enhanced and Time-aware Graph Convolutional Network (PTGCN) for the sequential recommendation. PTGCN models the sequential patterns and temporal dynamics between user-item interactions by defining a position-enhanced and time-aware graph convolution operation and learning the dynamic representations of users and items simultaneously on the bipartite graph with a self-attention aggregator. Also, it realizes the high-order connectivity between users and items by stacking multi-layer graph convolutions. To demonstrate the effectiveness of PTGCN, we carried out a comprehensive evaluation of PTGCN on three real-world datasets of different sizes compared with a few competitive baselines. Experimental results indicate that PTGCN outperforms several state-of-the-art models in terms of two commonly-used evaluation metrics for ranking.

arXiv:2107.02239공식 코드 · Papers with Code

Vision Xformers: Efficient Attention for Image Classification

Although transformers have become the neural architectures of choice for natural language processing, they require orders of magnitude more training data, GPU memory, and computations in order to compete with convolutional neural networks for computer vision. The attention mechanism of transformers scales quadratically with the length of the input sequence, and unrolled images have long sequence lengths. Plus, transformers lack an inductive bias that is appropriate for images. We tested three modifications to vision transformer (ViT) architectures that address these shortcomings. Firstly, we alleviate the quadratic bottleneck by using linear attention mechanisms, called X-formers (such that, X in {Performer, Linformer, Nystr\"omformer}), thereby creating Vision X-formers (ViXs). This resulted in up to a seven times reduction in the GPU memory requirement. We also compared their performance with FNet and multi-layer perceptron mixers, which further reduced the GPU memory requirement. Secondly, we introduced an inductive bias for images by replacing the initial linear embedding layer by convolutional layers in ViX, which significantly increased classification accuracy without increasing the model size. Thirdly, we replaced the learnable 1D position embeddings in ViT with Rotary Position Embedding (RoPE), which increases the classification accuracy for the same model size. We believe that incorporating such changes can democratize transformers by making them accessible to those with limited data and computing resources.

arXiv:2107.00761공식 코드 · Papers with Code

On the Bike Spreading Problem

A free-floating bike-sharing system (FFBSS) is a dockless rental system where an individual can borrow a bike and returns it anywhere, within the service area. To improve the rental service, available bikes should be distributed over the entire service area: a customer leaving from any position is then more likely to find a near bike and then to use the service. Moreover, spreading bikes among the entire service area increases urban spatial equity since the benefits of FFBSS are not a prerogative of just a few zones. For guaranteeing such distribution, the FFBSS operator can use vans to manually relocate bikes, but it incurs high economic and environmental costs. We propose a novel approach that exploits the existing bike flows generated by customers to distribute bikes. More specifically, by envisioning the problem as an Influence Maximization problem, we show that it is possible to position batches of bikes on a small number of zones, and then the daily use of FFBSS will efficiently spread these bikes on a large area. We show that detecting these zones is NP-complete, but there exists a simple and efficient $1-1/e$ approximation algorithm; our approach is then evaluated on a dataset of rides from the free-floating bike-sharing system of the city of Padova.

arXiv:2107.00962공식 코드 · Papers with Code

Brain over Brawn: Using a Stereo Camera to Detect, Track, and Intercept a Faster UAV by Reconstructing the Intruder's Trajectory

This paper presents our approach to intercepting a faster intruder UAV, inspired by the MBZIRC 2020 Challenge 1. By utilizing a priori knowledge of the shape of the intruder's trajectory, we can calculate an interception point. Target tracking is based on image processing by a YOLOv3 Tiny convolutional neural network, combined with depth calculation using a gimbal-mounted ZED Mini stereo camera. We use RGB and depth data from the camera, devising a noise-reducing histogram-filter to extract the target's 3D position. Obtained 3D measurements of target's position are used to calculate the position, orientation, and size of a figure-eight shaped trajectory, which we approximate using a Bernoulli lemniscate. Once the approximation is deemed sufficiently precise, as measured by the distance between observations and estimate, we calculate an interception point to position the interceptor UAV directly on the intruder's path. Our method, which we have significantly improved based on the experience gathered during the MBZIRC competition, has been validated in simulation and through field experiments. Our results confirm that we have developed an efficient, visual-perception module that can extract information describing the intruder UAV's motion with precision sufficient to support interception planning. In a majority of our simulated encounters, we can track and intercept a target that moves 30% faster than the interceptor. Corresponding tests in an unstructured environment yielded 9 out of 12 successful results.