arXiv subjects

Xinlei Chen

Publications and source records attributed to Xinlei Chen.

At least 127 records · Page 7Linked to original sources

arXiv2024

Deconstructing Denoising Diffusion Models for Self-Supervised Learning

In this study, we examine the representation learning abilities of Denoising Diffusion Models (DDM) that were originally purposed for image generation. Our philosophy is to deconstruct a DDM, gradually transforming it into a classical Denoising Autoencoder (DAE). This deconstructive procedure allows us to explore how various components of modern DDMs influence self-supervised representation learning. We observe that only a very few modern components are critical for learning good representations, while many others are nonessential. Our study ultimately arrives at an approach that is highly simplified and to a large extent resembles a classical DAE. We hope our study will rekindle interest in a family of classical methods within the realm of modern self-supervised learning.

cs.CV↗

arXiv2024

Learning to (Learn at Test Time)

We reformulate the problem of supervised learning as learning to learn with two nested loops (i.e. learning problems). The inner loop learns on each individual instance with self-supervision before final prediction. The outer loop learns the self-supervised task used by the inner loop, such that its final prediction improves. Our inner loop turns out to be equivalent to linear attention when the inner-loop learner is only a linear model, and to self-attention when it is a kernel estimator. For practical comparison with linear or self-attention layers, we replace each of them in a transformer with an inner loop, so our outer loop is equivalent to training the architecture. When each inner-loop learner is a neural network, our approach vastly outperforms transformers with linear attention on ImageNet from 224 x 224 raw pixels in both accuracy and FLOPs, while (regular) transformers cannot run.

cs.LG↗

arXiv2024

R-MAE: Regions Meet Masked Autoencoders

In this work, we explore regions as a potential visual analogue of words for self-supervised image representation learning. Inspired by Masked Autoencoding (MAE), a generative pre-training baseline, we propose masked region autoencoding to learn from groups of pixels or regions. Specifically, we design an architecture which efficiently addresses the one-to-many mapping between images and regions, while being highly effective especially with high-quality regions. When integrated with MAE, our approach (R-MAE) demonstrates consistent improvements across various pre-training datasets and downstream detection and segmentation benchmarks, with negligible computational overheads. Beyond the quantitative evaluation, our analysis indicates the models pre-trained with masked region autoencoding unlock the potential for interactive segmentation. The code is provided at https://github.com/facebookresearch/r-mae.

cs.CV↗

arXiv2023

Path Generation for Wheeled Robots Autonomous Navigation on Vegetated Terrain

Wheeled robot navigation has been widely used in urban environments, but little research has been conducted on its navigation in wild vegetation. External sensors (LiDAR, camera etc.) are often used to construct point cloud map of the surrounding environment, however, the supporting rigid ground used for travelling cannot be detected due to the occlusion of vegetation. This often causes unsafe or not smooth path during planning process. To address the drawback, we propose the PE-RRT* algorithm, which effectively combines a novel support plane estimation method and sampling algorithm to generate real-time feasible and safe path in vegetation environments. In order to accurately estimate the support plane, we combine external perception and proprioception, and use Multivariate Gaussian Processe Regression (MV-GPR) to estimate the terrain at the sampling nodes. We build a physical experimental platform and conduct experiments in different outdoor environments. Experimental results show that our method has high safety, robustness and generalization.

cs.RO↗

arXiv2023

The Pre-explosion Environments and The Progenitor of SN 2023ixf from the Hobby Eberly Telescope Dark Energy Experiment (HETDEX)

Supernova (SN) 2023ixf was discovered on May 19th, 2023. The host galaxy, M101, was observed by the Hobby Eberly Telescope Dark Energy Experiment (HETDEX) collaboration over the period April 30, 2020 -- July 10, 2020, using the Visible Integral-field Replicable Unit Spectrograph (VIRUS; $3470\lesssimλ\lesssim5540$ Å) on the 10-m Hobby-Eberly Telescope (HET). The fiber filling factor within $\pm$ 30 arcsec of SN 2023ixf is 80% with a spatial resolution of 1 arcsec. The r<5.5 arcsec surroundings are 100% covered. This allows us to analyze the spatially resolved pre-explosion local environments of SN 2023ixf with nebular emission lines. The 2-dimensional (2D) maps of the extinction and the star-formation rate (SFR) surface density ($Σ_{\rm SFR}$) show weak increasing trends in the radial distributions within the r<5.5 arcsec regions, suggesting lower values of extinction and SFR in the vicinity of the progenitor of SN 2023ixf. The median extinction and that of the surface density of SFR within r<3 arcsec are $E(B-V)=0.06\pm0.14$, and $Σ_{\rm SFR}=10^{-5.44\pm0.66}~\rm M_{\odot}\cdot yr^{-1}\cdot arcsec^{-2}$. There is no significant change in extinction before and after the explosion. The gas metallicity does not change significantly with the separation from SN 2023ixf. The metal-rich branch of the $R_{23}$ calculations indicates that the gas metallicity around SN 2023ixf is similar to the solar metallicity ($\sim Z_{\odot}$). The archival deep images from the Canada-France-Hawaii Telescope Legacy Survey (CFHTLS) show a clear detection of the progenitor of SN 2023ixf in the $z$-band at $22.778\pm0.063$ mag, but non-detections in the remaining four bands of CFHTLS ($u,g,r,i$). The results suggest a massive progenitor of $\approx$ 22 $M_\odot$.

astro-ph.GA↗

arXiv2023

Searching for new globular clusters in M 31 with Gaia EDR3

We found 50 new globular cluster (GC) candidates around M\,31 with Gaia Early Data Release 3 (EDR3), with the help from Pan-STARRS1 DR1 magnitudes and Pan-Andromeda Archaeological Survey (PAndAS) images. Based on the latest Revised Bologna Catalog and \textit{simbad}, we trained 2 Random Forest (RF) classifiers, the first one to distinguish extended sources from point sources and the second one to further select GCs from extended sources. From 1.85 million sources of $16^m{<}g{<}19.5^m$ and within a large area of $\sim$392\,deg$^2$ around M\,31, we selected 20,658 extended sources and 1,934 initial GC candidates. After visual inspection of the PAndAS images to eliminate the contamination of non-cluster sources, particularly galaxies, we finally got 50 candidates. These candidates are divided into 3 types (\textbf{a}, \textbf{b}, \textbf{c}) according to their projected distance $D$ to the center of M\,31 and their probability to be a true GC, $P_{GC}$, which is calculated by our second RF classifier. Among these candidates, 14 are found to be associated (in projection) with the large-scale structures within the halo of M\,31. We also provided several simple parameter criteria for selecting extended sources effectively from the Gaia EDR3, which can reach a completeness of 92.1\% with a contamination fraction lower than 10\%.

astro-ph.GA↗

arXiv2023

Improving Selective Visual Question Answering by Learning from Your Peers

Despite advances in Visual Question Answering (VQA), the ability of models to assess their own correctness remains underexplored. Recent work has shown that VQA models, out-of-the-box, can have difficulties abstaining from answering when they are wrong. The option to abstain, also called Selective Prediction, is highly relevant when deploying systems to users who must trust the system's output (e.g., VQA assistants for users with visual impairments). For such scenarios, abstention can be especially important as users may provide out-of-distribution (OOD) or adversarial inputs that make incorrect answers more likely. In this work, we explore Selective VQA in both in-distribution (ID) and OOD scenarios, where models are presented with mixtures of ID and OOD data. The goal is to maximize the number of questions answered while minimizing the risk of error on those questions. We propose a simple yet effective Learning from Your Peers (LYP) approach for training multimodal selection functions for making abstention decisions. Our approach uses predictions from models trained on distinct subsets of the training data as targets for optimizing a Selective VQA model. It does not require additional manual labels or held-out data and provides a signal for identifying examples that are easy/difficult to generalize to. In our extensive evaluations, we show this benefits a number of models across different architectures and scales. Overall, for ID, we reach 32.92% in the selective prediction metric coverage at 1% risk of error (C@1%) which doubles the previous best coverage of 15.79% on this task. For mixed ID/OOD, using models' softmax confidences for abstention decisions performs very poorly, answering <5% of questions at 1% risk of error even when faced with only 10% OOD examples, but a learned selection function with LYP can increase that to 25.38% C@1%.

cs.CV↗

arXiv2023

ConvNeXt V2: Co-designing and Scaling ConvNets with Masked Autoencoders

Driven by improved architectures and better representation learning frameworks, the field of visual recognition has enjoyed rapid modernization and performance boost in the early 2020s. For example, modern ConvNets, represented by ConvNeXt, have demonstrated strong performance in various scenarios. While these models were originally designed for supervised learning with ImageNet labels, they can also potentially benefit from self-supervised learning techniques such as masked autoencoders (MAE). However, we found that simply combining these two approaches leads to subpar performance. In this paper, we propose a fully convolutional masked autoencoder framework and a new Global Response Normalization (GRN) layer that can be added to the ConvNeXt architecture to enhance inter-channel feature competition. This co-design of self-supervised learning techniques and architectural improvement results in a new model family called ConvNeXt V2, which significantly improves the performance of pure ConvNets on various recognition benchmarks, including ImageNet classification, COCO detection, and ADE20K segmentation. We also provide pre-trained ConvNeXt V2 models of various sizes, ranging from an efficient 3.7M-parameter Atto model with 76.7% top-1 accuracy on ImageNet, to a 650M Huge model that achieves a state-of-the-art 88.9% accuracy using only public training data.

cs.CV↗

arXiv2022

Towards Efficient Communications in Federated Learning: A Contemporary Survey

In the traditional distributed machine learning scenario, the user's private data is transmitted between clients and a central server, which results in significant potential privacy risks. In order to balance the issues of data privacy and joint training of models, federated learning (FL) is proposed as a particular distributed machine learning procedure with privacy protection mechanisms, which can achieve multi-party collaborative computing without revealing the original data. However, in practice, FL faces a variety of challenging communication problems. This review seeks to elucidate the relationship between these communication issues by methodically assessing the development of FL communication research from three perspectives: communication efficiency, communication environment, and communication resource allocation. Firstly, we sort out the current challenges existing in the communications of FL. Second, we have collated FL communications-related papers and described the overall development trend of the field based on their logical relationship. Ultimately, we discuss the future directions of research for communications in FL.

cs.DC↗

arXiv2022

UniT3D: A Unified Transformer for 3D Dense Captioning and Visual Grounding

Performing 3D dense captioning and visual grounding requires a common and shared understanding of the underlying multimodal relationships. However, despite some previous attempts on connecting these two related tasks with highly task-specific neural modules, it remains understudied how to explicitly depict their shared nature to learn them simultaneously. In this work, we propose UniT3D, a simple yet effective fully unified transformer-based architecture for jointly solving 3D visual grounding and dense captioning. UniT3D enables learning a strong multimodal representation across the two tasks through a supervised joint pre-training scheme with bidirectional and seq-to-seq objectives. With a generic architecture design, UniT3D allows expanding the pre-training scope to more various training sources such as the synthesized data from 2D prior knowledge to benefit 3D vision-language tasks. Extensive experiments and analysis demonstrate that UniT3D obtains significant gains for 3D dense captioning and visual grounding.

cs.CV↗

arXiv2022

EurNet: Efficient Multi-Range Relational Modeling of Spatial Multi-Relational Data

Modeling spatial relationship in the data remains critical across many different tasks, such as image classification, semantic segmentation and protein structure understanding. Previous works often use a unified solution like relative positional encoding. However, there exists different kinds of spatial relations, including short-range, medium-range and long-range relations, and modeling them separately can better capture the focus of different tasks on the multi-range relations (e.g., short-range relations can be important in instance segmentation, while long-range relations should be upweighted for semantic segmentation). In this work, we introduce the EurNet for Efficient multi-range relational modeling. EurNet constructs the multi-relational graph, where each type of edge corresponds to short-, medium- or long-range spatial interactions. In the constructed graph, EurNet adopts a novel modeling layer, called gated relational message passing (GRMP), to propagate multi-relational information across the data. GRMP captures multiple relations within the data with little extra computational cost. We study EurNets in two important domains for image and protein structure modeling. Extensive experiments on ImageNet classification, COCO object detection and ADE20K semantic segmentation verify the gains of EurNet over the previous SoTA FocalNet. On the EC and GO protein function prediction benchmarks, EurNet consistently surpasses the previous SoTA GearNet. Our results demonstrate the strength of EurNets on modeling spatial multi-relational data from various domains. The implementations of EurNet for image modeling are available at https://github.com/hirl-team/EurNet-Image . The implementations for other applied domains/tasks will be released soon.

cs.LG↗

arXiv2022

Exploring Long-Sequence Masked Autoencoders

Masked Autoencoding (MAE) has emerged as an effective approach for pre-training representations across multiple domains. In contrast to discrete tokens in natural languages, the input for image MAE is continuous and subject to additional specifications. We systematically study each input specification during the pre-training stage, and find sequence length is a key axis that further scales MAE. Our study leads to a long-sequence version of MAE with minimal changes to the original recipe, by just decoupling the mask size from the patch size. For object detection and semantic segmentation, our long-sequence MAE shows consistent gains across all the experimental setups without extra computation cost during the transfer. While long-sequence pre-training is discerned most beneficial for detection and segmentation, we also achieve strong results on ImageNet-1K classification by keeping a standard image size and only increasing the sequence length. We hope our findings can provide new insights and avenues for scaling in computer vision.

cs.CV↗

arXiv2022

Towards Demystifying Representation Learning with Non-contrastive Self-supervision

Non-contrastive methods of self-supervised learning (such as BYOL and SimSiam) learn representations by minimizing the distance between two views of the same image. These approaches have achieved remarkable performance in practice, but the theoretical understanding lags behind. Tian et al. 2021 explained why the representation does not collapse to zero, however, how the feature is learned still remains mysterious. In our work, we prove in a linear network, non-contrastive methods learn a desirable projection matrix and also reduce the sample complexity on downstream tasks. Our analysis suggests that weight decay acts as an implicit threshold that discards the features with high variance under data augmentations, and keeps the features with low variance. Inspired by our theory, we design a simpler and more computationally efficient algorithm DirectCopy by removing the eigen-decomposition step in the original DirectPred algorithm in Tian et al. 2021. Our experiments show that DirectCopy rivals or even outperforms DirectPred on STL-10, CIFAR-10, CIFAR-100, and ImageNet.

cs.LG↗

arXiv2022

Test-Time Training with Masked Autoencoders

Test-time training adapts to a new test distribution on the fly by optimizing a model for each test input using self-supervision. In this paper, we use masked autoencoders for this one-sample learning problem. Empirically, our simple method improves generalization on many visual benchmarks for distribution shifts. Theoretically, we characterize this improvement in terms of the bias-variance trade-off.

cs.CV↗

arXiv2022

Point-Level Region Contrast for Object Detection Pre-Training

In this work we present point-level region contrast, a self-supervised pre-training approach for the task of object detection. This approach is motivated by the two key factors in detection: localization and recognition. While accurate localization favors models that operate at the pixel- or point-level, correct recognition typically relies on a more holistic, region-level view of objects. Incorporating this perspective in pre-training, our approach performs contrastive learning by directly sampling individual point pairs from different regions. Compared to an aggregated representation per region, our approach is more robust to the change in input region quality, and further enables us to implicitly improve initial region assignments via online knowledge distillation during training. Both advantages are important when dealing with imperfect regions encountered in the unsupervised setting. Experiments show point-level region contrast improves on state-of-the-art pre-training methods for object detection and segmentation across multiple tasks and datasets, and we provide extensive ablation studies and visualizations to aid understanding. Code will be made available.

cs.CV↗

arXiv2022

On the Importance of Asymmetry for Siamese Representation Learning

Many recent self-supervised frameworks for visual representation learning are based on certain forms of Siamese networks. Such networks are conceptually symmetric with two parallel encoders, but often practically asymmetric as numerous mechanisms are devised to break the symmetry. In this work, we conduct a formal study on the importance of asymmetry by explicitly distinguishing the two encoders within the network -- one produces source encodings and the other targets. Our key insight is keeping a relatively lower variance in target than source generally benefits learning. This is empirically justified by our results from five case studies covering different variance-oriented designs, and is aligned with our preliminary theoretical analysis on the baseline. Moreover, we find the improvements from asymmetric designs generalize well to longer training schedules, multiple other frameworks and newer backbones. Finally, the combined effect of several asymmetric designs achieves a state-of-the-art accuracy on ImageNet linear probing and competitive results on downstream transfer. We hope our exploration will inspire more research in exploiting asymmetry for Siamese representation learning.

cs.CV↗

arXiv2022

LoopITR: Combining Dual and Cross Encoder Architectures for Image-Text Retrieval

Dual encoders and cross encoders have been widely used for image-text retrieval. Between the two, the dual encoder encodes the image and text independently followed by a dot product, while the cross encoder jointly feeds image and text as the input and performs dense multi-modal fusion. These two architectures are typically modeled separately without interaction. In this work, we propose LoopITR, which combines them in the same network for joint learning. Specifically, we let the dual encoder provide hard negatives to the cross encoder, and use the more discriminative cross encoder to distill its predictions back to the dual encoder. Both steps are efficiently performed together in the same model. Our work centers on empirical analyses of this combined architecture, putting the main focus on the design of the distillation objective. Our experimental results highlight the benefits of training the two encoders in the same network, and demonstrate that distillation can be quite effective with just a few hard negative examples. Experiments on two standard datasets (Flickr30K and COCO) show our approach achieves state-of-the-art dual encoder performance when compared with approaches using a similar amount of data.

cs.CV↗

arXiv2021

Masked Autoencoders Are Scalable Vision Learners

This paper shows that masked autoencoders (MAE) are scalable self-supervised learners for computer vision. Our MAE approach is simple: we mask random patches of the input image and reconstruct the missing pixels. It is based on two core designs. First, we develop an asymmetric encoder-decoder architecture, with an encoder that operates only on the visible subset of patches (without mask tokens), along with a lightweight decoder that reconstructs the original image from the latent representation and mask tokens. Second, we find that masking a high proportion of the input image, e.g., 75%, yields a nontrivial and meaningful self-supervisory task. Coupling these two designs enables us to train large models efficiently and effectively: we accelerate training (by 3x or more) and improve accuracy. Our scalable approach allows for learning high-capacity models that generalize well: e.g., a vanilla ViT-Huge model achieves the best accuracy (87.8%) among methods that use only ImageNet-1K data. Transfer performance in downstream tasks outperforms supervised pre-training and shows promising scaling behavior.

cs.CV↗