PyTorchKR Papers

최신 논문

총 715편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2304.13567공식 코드 · Papers with Code

Technical Report: Impact of Position Bias on Language Models in Token Classification

Language Models (LMs) have shown state-of-the-art performance in Natural Language Processing (NLP) tasks. Downstream tasks such as Named Entity Recognition (NER) or Part-of-Speech (POS) tagging are known to suffer from data imbalance issues, particularly regarding the ratio of positive to negative examples and class disparities. This paper investigates an often-overlooked issue of encoder models, specifically the position bias of positive examples in token classification tasks. For completeness, we also include decoders in the evaluation. We evaluate the impact of position bias using different position embedding techniques, focusing on BERT with Absolute Position Embedding (APE), Relative Position Embedding (RPE), and Rotary Position Embedding (RoPE). Therefore, we conduct an in-depth evaluation of the impact of position bias on the performance of LMs when fine-tuned on token classification benchmarks. Our study includes CoNLL03 and OntoNote5.0 for NER, English Tree Bank UD\_en, and TweeBank for POS tagging. We propose an evaluation approach to investigate position bias in transformer models. We show that LMs can suffer from this bias with an average drop ranging from 3\% to 9\% in their performance. To mitigate this effect, we propose two methods: Random Position Shifting and Context Perturbation, that we apply on batches during the training process. The results show an improvement of $\approx$ 2\% in the performance of the model on CoNLL03, UD\_en, and TweeBank.

arXiv:2304.09516공식 코드 · Papers with Code

Controlling keywords and their positions in text generation

One of the challenges in text generation is to control text generation as intended by the user. Previous studies proposed specifying the keywords that should be included in the generated text. However, this approach is insufficient to generate text that reflect the user's intent. For example, placing an important keyword at the beginning of the text would help attract the reader's attention; however, existing methods do not enable such flexible control. In this paper, we tackle a novel task of controlling not only keywords but also the position of each keyword in the text generation. To this end, we propose a task-independent method that uses special tokens to control the relative position of keywords. Experimental results on summarization and story generation tasks show that the proposed method can control keywords and their positions. The experimental results also demonstrate that controlling the keyword positions can generate summary texts that are closer to the user's intent than baseline.

arXiv:2304.08571공식 코드 · Papers with Code

Optimal Robust Network Design: Formulations and Algorithms for Maximizing Algebraic Connectivity

This paper focuses on designing edge-weighted networks, whose robustness is characterized by maximizing algebraic connectivity, or the second smallest eigenvalue of the Laplacian matrix. This problem is motivated by cooperative vehicle localization, where accurately estimating relative position measurements and establishing communication links are essential. We also examine an associated problem where every robot is limited by payload, budget, and communication to pick no more than a specified number of relative position measurements. The basic underlying formulation for these problems is nonlinear and is known to be NP-hard. Our approach formulates this problem as a Mixed Integer Semi-Definite Program (MISDP), later reformulated into a Mixed Integer Linear Program (MILP) for obtaining optimal solutions using cutting plane algorithms. We introduce a novel upper-bounding algorithm based on principal minor characterization of positive semi-definite matrices and discuss a degree-constrained lower bounding formulation inspired by robust network structures. In addition, we propose a maximum cost heuristic with low computational complexity to identify high-quality feasible solutions for instances involving up to one hundred nodes. We show extensive computational results corroborating our proposed methods.

arXiv:2304.03945공식 코드 · Papers with Code

Knowledge Relation Rank Enhanced Heterogeneous Learning Interaction Modeling for Neural Graph Forgetting Knowledge Tracing

Recently, knowledge tracing models have been applied in educational data mining such as the Self-attention knowledge tracing model(SAKT), which models the relationship between exercises and Knowledge concepts(Kcs). However, relation modeling in traditional Knowledge tracing models only considers the static question-knowledge relationship and knowledge-knowledge relationship and treats these relationships with equal importance. This kind of relation modeling is difficult to avoid the influence of subjective labeling and considers the relationship between exercises and KCs, or KCs and KCs separately. In this work, a novel knowledge tracing model, named Knowledge Relation Rank Enhanced Heterogeneous Learning Interaction Modeling for Neural Graph Forgetting Knowledge Tracing(NGFKT), is proposed to reduce the impact of the subjective labeling by calibrating the skill relation matrix and the Q-matrix and apply the Graph Convolutional Network(GCN) to model the heterogeneous interactions between students, exercises, and skills. Specifically, the skill relation matrix and Q-matrix are generated by the Knowledge Relation Importance Rank Calibration method(KRIRC). Then the calibrated skill relation matrix, Q-matrix, and the heterogeneous interactions are treated as the input of the GCN to generate the exercise embedding and skill embedding. Next, the exercise embedding, skill embedding, item difficulty, and contingency table are incorporated to generate an exercise relation matrix as the inputs of the Position-Relation-Forgetting attention mechanism. Finally, the Position-Relation-Forgetting attention mechanism is applied to make the predictions. Experiments are conducted on the two public educational datasets and results indicate that the NGFKT model outperforms all baseline models in terms of AUC, ACC, and Performance Stability(PS).

arXiv:2304.04742공식 코드 · Papers with Code

Detection Transformer with Stable Matching

This paper is concerned with the matching stability problem across different decoder layers in DEtection TRansformers (DETR). We point out that the unstable matching in DETR is caused by a multi-optimization path problem, which is highlighted by the one-to-one matching design in DETR. To address this problem, we show that the most important design is to use and only use positional metrics (like IOU) to supervise classification scores of positive examples. Under the principle, we propose two simple yet effective modifications by integrating positional metrics to DETR's classification loss and matching cost, named position-supervised loss and position-modulated cost. We verify our methods on several DETR variants. Our methods show consistent improvements over baselines. By integrating our methods with DINO, we achieve 50.4 and 51.5 AP on the COCO detection benchmark using ResNet-50 backbones under 12 epochs and 24 epochs training settings, achieving a new record under the same setting. We achieve 63.8 AP on COCO detection test-dev with a Swin-Large backbone. Our code will be made available at https://github.com/IDEA-Research/Stable-DINO.

arXiv:2303.18248공식 코드 · Papers with Code

Towards Flexible Multi-modal Document Models

Creative workflows for generating graphical documents involve complex inter-related tasks, such as aligning elements, choosing appropriate fonts, or employing aesthetically harmonious colors. In this work, we attempt at building a holistic model that can jointly solve many different design tasks. Our model, which we denote by FlexDM, treats vector graphic documents as a set of multi-modal elements, and learns to predict masked fields such as element type, position, styling attributes, image, or text, using a unified architecture. Through the use of explicit multi-task learning and in-domain pre-training, our model can better capture the multi-modal relationships among the different document fields. Experimental results corroborate that our single FlexDM is able to successfully solve a multitude of different design tasks, while achieving performance that is competitive with task-specific and costly baselines.

arXiv:2303.18181공식 코드 · Papers with Code

A Closer Look at Parameter-Efficient Tuning in Diffusion Models

Large-scale diffusion models like Stable Diffusion are powerful and find various real-world applications while customizing such models by fine-tuning is both memory and time inefficient. Motivated by the recent progress in natural language processing, we investigate parameter-efficient tuning in large diffusion models by inserting small learnable modules (termed adapters). In particular, we decompose the design space of adapters into orthogonal factors -- the input position, the output position as well as the function form, and perform Analysis of Variance (ANOVA), a classical statistical approach for analyzing the correlation between discrete (design options) and continuous variables (evaluation metrics). Our analysis suggests that the input position of adapters is the critical factor influencing the performance of downstream tasks. Then, we carefully study the choice of the input position, and we find that putting the input position after the cross-attention block can lead to the best performance, validated by additional visualization analyses. Finally, we provide a recipe for parameter-efficient tuning in diffusion models, which is comparable if not superior to the fully fine-tuned baseline (e.g., DreamBooth) with only 0.75 \% extra parameters, across various customized tasks.

arXiv:2303.16884공식 코드 · Papers with Code

Instant Photorealistic Neural Radiance Fields Stylization

We present Instant Neural Radiance Fields Stylization, a novel approach for multi-view image stylization for the 3D scene. Our approach models a neural radiance field based on neural graphics primitives, which use a hash table-based position encoder for position embedding. We split the position encoder into two parts, the content and style sub-branches, and train the network for normal novel view image synthesis with the content and style targets. In the inference stage, we execute AdaIN to the output features of the position encoder, with content and style voxel grid features as reference. With the adjusted features, the stylization of novel view images could be obtained. Our method extends the style target from style images to image sets of scenes and does not require additional network training for stylization. Given a set of images of 3D scenes and a style target(a style image or another set of 3D scenes), our method can generate stylized novel views with a consistent appearance at various view angles in less than 10 minutes on modern GPU hardware. Extensive experimental results demonstrate the validity and superiority of our method.

arXiv:2303.15936공식 코드 · Papers with Code

Searching for long faint astronomical high energy transients: a data driven approach

HERMES (High Energy Rapid Modular Ensemble of Satellites) pathfinder is an in-orbit demonstration consisting of a constellation of six 3U nano-satellites hosting simple but innovative detectors for the monitoring of cosmic high-energy transients. The main objective of HERMES Pathfinder is to prove that accurate position of high-energy cosmic transients can be obtained using miniaturized hardware. The transient position is obtained by studying the delay time of arrival of the signal to different detectors hosted by nano-satellites on low Earth orbits. To this purpose, the goal is to achive an overall accuracy of a fraction of a micro-second. In this context, we need to develop novel tools to fully exploit the future scientific data output of HERMES Pathfinder. In this paper, we introduce a new framework to assess the background count rate of a space-born, high energy detector; a key step towards the identification of faint astrophysical transients. We employ a Neural Network (NN) to estimate the background lightcurves on different timescales. Subsequently, we employ a fast change-point and anomaly detection technique to isolate observation segments where statistically significant excesses in the observed count rate relative to the background estimate exist. We test the new software on archival data from the NASA Fermi Gamma-ray Burst Monitor (GBM), which has a collecting area and background level of the same order of magnitude to those of HERMES Pathfinder. The NN performances are discussed and analyzed over period of both high and low solar activity. We were able to confirm events in the Fermi/GBM catalog and found events, not present in Fermi/GBM database, that could be attributed to Solar Flares, Terrestrial Gamma-ray Flashes, Gamma-Ray Bursts, Galactic X-ray flash. Seven of these are selected and analyzed further, providing an estimate of localisation and a tentative classification.

arXiv:2303.13509공식 코드 · Papers with Code

Position-Guided Point Cloud Panoptic Segmentation Transformer

DEtection TRansformer (DETR) started a trend that uses a group of learnable queries for unified visual perception. This work begins by applying this appealing paradigm to LiDAR-based point cloud segmentation and obtains a simple yet effective baseline. Although the naive adaptation obtains fair results, the instance segmentation performance is noticeably inferior to previous works. By diving into the details, we observe that instances in the sparse point clouds are relatively small to the whole scene and often have similar geometry but lack distinctive appearance for segmentation, which are rare in the image domain. Considering instances in 3D are more featured by their positional information, we emphasize their roles during the modeling and design a robust Mixed-parameterized Positional Embedding (MPE) to guide the segmentation process. It is embedded into backbone features and later guides the mask prediction and query update processes iteratively, leading to Position-Aware Segmentation (PA-Seg) and Masked Focal Attention (MFA). All these designs impel the queries to attend to specific regions and identify various instances. The method, named Position-guided Point cloud Panoptic segmentation transFormer (P3Former), outperforms previous state-of-the-art methods by 3.4% and 1.2% PQ on SemanticKITTI and nuScenes benchmark, respectively. The source code and models are available at https://github.com/SmartBot-PJLab/P3Former .

arXiv:2303.12054공식 코드 · Papers with Code

Influencer Backdoor Attack on Semantic Segmentation

When a small number of poisoned samples are injected into the training dataset of a deep neural network, the network can be induced to exhibit malicious behavior during inferences, which poses potential threats to real-world applications. While they have been intensively studied in classification, backdoor attacks on semantic segmentation have been largely overlooked. Unlike classification, semantic segmentation aims to classify every pixel within a given image. In this work, we explore backdoor attacks on segmentation models to misclassify all pixels of a victim class by injecting a specific trigger on non-victim pixels during inferences, which is dubbed Influencer Backdoor Attack (IBA). IBA is expected to maintain the classification accuracy of non-victim pixels and mislead classifications of all victim pixels in every single inference and could be easily applied to real-world scenes. Based on the context aggregation ability of segmentation models, we proposed a simple, yet effective, Nearest-Neighbor trigger injection strategy. We also introduce an innovative Pixel Random Labeling strategy which maintains optimal performance even when the trigger is placed far from the victim pixels. Our extensive experiments reveal that current segmentation models do suffer from backdoor attacks, demonstrate IBA real-world applicability, and show that our proposed techniques can further increase attack performance.

arXiv:2303.10209공식 코드 · Papers with Code

CAPE: Camera View Position Embedding for Multi-View 3D Object Detection

In this paper, we address the problem of detecting 3D objects from multi-view images. Current query-based methods rely on global 3D position embeddings (PE) to learn the geometric correspondence between images and 3D space. We claim that directly interacting 2D image features with global 3D PE could increase the difficulty of learning view transformation due to the variation of camera extrinsics. Thus we propose a novel method based on CAmera view Position Embedding, called CAPE. We form the 3D position embeddings under the local camera-view coordinate system instead of the global coordinate system, such that 3D position embedding is free of encoding camera extrinsic parameters. Furthermore, we extend our CAPE to temporal modeling by exploiting the object queries of previous frames and encoding the ego-motion for boosting 3D object detection. CAPE achieves state-of-the-art performance (61.0% NDS and 52.5% mAP) among all LiDAR-free methods on nuScenes dataset. Codes and models are available on \href{https://github.com/PaddlePaddle/Paddle3D}{Paddle3D} and \href{https://github.com/kaixinbear/CAPE}{PyTorch Implementation}.

arXiv:2303.08137공식 코드 · Papers with Code

LayoutDM: Discrete Diffusion Model for Controllable Layout Generation

Controllable layout generation aims at synthesizing plausible arrangement of element bounding boxes with optional constraints, such as type or position of a specific element. In this work, we try to solve a broad range of layout generation tasks in a single model that is based on discrete state-space diffusion models. Our model, named LayoutDM, naturally handles the structured layout data in the discrete representation and learns to progressively infer a noiseless layout from the initial input, where we model the layout corruption process by modality-wise discrete diffusion. For conditional generation, we propose to inject layout constraints in the form of masking or logit adjustment during inference. We show in the experiments that our LayoutDM successfully generates high-quality layouts and outperforms both task-specific and task-agnostic baselines on several layout tasks.

arXiv:2303.05269공식 코드 · Papers with Code

Effective Pseudo-Labeling based on Heatmap for Unsupervised Domain Adaptation in Cell Detection

Cell detection is an important task in biomedical research. Recently, deep learning methods have made it possible to improve the performance of cell detection. However, a detection network trained with training data under a specific condition (source domain) may not work well on data under other conditions (target domains), which is called the domain shift problem. In particular, cells are cultured under different conditions depending on the purpose of the research. Characteristics, e.g., the shapes and density of the cells, change depending on the conditions, and such changes may cause domain shift problems. Here, we propose an unsupervised domain adaptation method for cell detection using a pseudo-cell-position heatmap, where the cell centroid is at the peak of a Gaussian distribution in the map and selective pseudo-labeling. In the prediction result for the target domain, even if the peak location is correct, the signal distribution around the peak often has a non-Gaussian shape. The pseudo-cell-position heatmap is thus re-generated using the peak positions in the predicted heatmap to have a clear Gaussian shape. Our method selects confident pseudo-cell-position heatmaps based on uncertainty and curriculum learning. We conducted numerous experiments showing that, compared with the existing methods, our method improved detection performance under different conditions.

arXiv:2303.04346공식 코드 · Papers with Code

Semi-Supervised 2D Human Pose Estimation Driven by Position Inconsistency Pseudo Label Correction Module

In this paper, we delve into semi-supervised 2D human pose estimation. The previous method ignored two problems: (i) When conducting interactive training between large model and lightweight model, the pseudo label of lightweight model will be used to guide large models. (ii) The negative impact of noise pseudo labels on training. Moreover, the labels used for 2D human pose estimation are relatively complex: keypoint category and keypoint position. To solve the problems mentioned above, we propose a semi-supervised 2D human pose estimation framework driven by a position inconsistency pseudo label correction module (SSPCM). We introduce an additional auxiliary teacher and use the pseudo labels generated by the two teacher model in different periods to calculate the inconsistency score and remove outliers. Then, the two teacher models are updated through interactive training, and the student model is updated using the pseudo labels generated by two teachers. To further improve the performance of the student model, we use the semi-supervised Cut-Occlude based on pseudo keypoint perception to generate more hard and effective samples. In addition, we also proposed a new indoor overhead fisheye human keypoint dataset WEPDTOF-Pose. Extensive experiments demonstrate that our method outperforms the previous best semi-supervised 2D human pose estimation method. We will release the code and dataset at https://github.com/hlz0606/SSPCM.

arXiv:2303.03595공식 코드 · Papers with Code

LoGoNet: Towards Accurate 3D Object Detection with Local-to-Global Cross-Modal Fusion

LiDAR-camera fusion methods have shown impressive performance in 3D object detection. Recent advanced multi-modal methods mainly perform global fusion, where image features and point cloud features are fused across the whole scene. Such practice lacks fine-grained region-level information, yielding suboptimal fusion performance. In this paper, we present the novel Local-to-Global fusion network (LoGoNet), which performs LiDAR-camera fusion at both local and global levels. Concretely, the Global Fusion (GoF) of LoGoNet is built upon previous literature, while we exclusively use point centroids to more precisely represent the position of voxel features, thus achieving better cross-modal alignment. As to the Local Fusion (LoF), we first divide each proposal into uniform grids and then project these grid centers to the images. The image features around the projected grid points are sampled to be fused with position-decorated point cloud features, maximally utilizing the rich contextual information around the proposals. The Feature Dynamic Aggregation (FDA) module is further proposed to achieve information interaction between these locally and globally fused features, thus producing more informative multi-modal features. Extensive experiments on both Waymo Open Dataset (WOD) and KITTI datasets show that LoGoNet outperforms all state-of-the-art 3D detection methods. Notably, LoGoNet ranks 1st on Waymo 3D object detection leaderboard and obtains 81.02 mAPH (L2) detection performance. It is noteworthy that, for the first time, the detection performance on three classes surpasses 80 APH (L2) simultaneously. Code will be available at \url{https://github.com/sankin97/LoGoNet}.

arXiv:2303.02909공식 코드 · Papers with Code

Dynamic Prompting: A Unified Framework for Prompt Tuning

It has been demonstrated that the art of prompt tuning is highly effective in efficiently extracting knowledge from pretrained foundation models, encompassing pretrained language models (PLMs), vision pretrained models, and vision-language (V-L) models. However, the efficacy of employing fixed soft prompts with a predetermined position for concatenation with inputs for all instances, irrespective of their inherent disparities, remains uncertain. Variables such as the position, length, and representations of prompts across diverse instances and tasks can substantially influence the performance of prompt tuning. In this context, we provide a theoretical analysis, which reveals that optimizing the position of the prompt to encompass the input can capture additional semantic information that traditional prefix or postfix prompt tuning methods fail to capture. Building upon our analysis, we present a unified dynamic prompt (DP) tuning strategy that dynamically determines different factors of prompts based on specific tasks and instances. To accomplish this, we employ a lightweight learning network with Gumble-Softmax, allowing us to learn instance-dependent guidance. Experimental results underscore the significant performance improvement achieved by dynamic prompt tuning across a wide range of tasks, including NLP tasks, vision recognition tasks, and vision-language tasks. Furthermore, we establish the universal applicability of our approach under full-data, few-shot, and multitask scenarios. Codes are available at https://github.com/Xianjun-Yang/DPT.

arXiv:2303.01795공식 코드 · Papers with Code

PAGE: A Position-Aware Graph-Based Model for Emotion Cause Entailment in Conversation

Conversational Causal Emotion Entailment (C2E2) is a task that aims at recognizing the causes corresponding to a target emotion in a conversation. The order of utterances in the conversation affects the causal inference. However, most current position encoding strategies ignore the order relation among utterances and speakers. To address the issue, we devise a novel position-aware graph to encode the entire conversation, fully modeling causal relations among utterances. The comprehensive experiments show that our method consistently achieves state-of-the-art performance on two challenging test sets, proving the effectiveness of our model. Our source code is available on Github: https://github.com/XiaojieGu/PAGE.

arXiv:2303.01751공식 코드 · Papers with Code

Deep Momentum Multi-Marginal Schrödinger Bridge

It is a crucial challenge to reconstruct population dynamics using unlabeled samples from distributions at coarse time intervals. Recent approaches such as flow-based models or Schr\"odinger Bridge (SB) models have demonstrated appealing performance, yet the inferred sample trajectories either fail to account for the underlying stochasticity or are $\underline{D}$eep $\underline{M}$omentum Multi-Marginal $\underline{S}$chr\"odinger $\underline{B}$ridge(DMSB), a novel computational framework that learns the smooth measure-valued spline for stochastic systems that satisfy position marginal constraints across time. By tailoring the celebrated Bregman Iteration and extending the Iteration Proportional Fitting to phase space, we manage to handle high-dimensional multi-marginal trajectory inference tasks efficiently. Our algorithm outperforms baselines significantly, as evidenced by experiments for synthetic datasets and a real-world single-cell RNA sequence dataset. Additionally, the proposed approach can reasonably reconstruct the evolution of velocity distribution, from position snapshots only, when there is a ground truth velocity that is nevertheless inaccessible.

arXiv:2303.01491공식 코드 · Papers with Code

Transferring Models Trained on Natural Images to 3D MRI via Position Encoded Slice Models

Transfer learning has remarkably improved computer vision. These advances also promise improvements in neuroimaging, where training set sizes are often small. However, various difficulties arise in directly applying models pretrained on natural images to radiologic images, such as MRIs. In particular, a mismatch in the input space (2D images vs. 3D MRIs) restricts the direct transfer of models, often forcing us to consider only a few MRI slices as input. To this end, we leverage the 2D-Slice-CNN architecture of Gupta et al. (2021), which embeds all the MRI slices with 2D encoders (neural networks that take 2D image input) and combines them via permutation-invariant layers. With the insight that the pretrained model can serve as the 2D encoder, we initialize the 2D encoder with ImageNet pretrained weights that outperform those initialized and trained from scratch on two neuroimaging tasks -- brain age prediction on the UK Biobank dataset and Alzheimer's disease detection on the ADNI dataset. Further, we improve the modeling capabilities of 2D-Slice models by incorporating spatial information through position embeddings, which can improve the performance in some cases.