Search

The Segment Anything Model (SAM) is the first foundation model for general image segmentation. It has achieved impressive results on various natural image segmentation tasks. However, medical image segmentation (MIS) is more challenging because of the complex modalities, fine anatomical structures, uncertain and complex object boundaries, and wide-range object scales. To fully validate SAM's performance on medical data, we collected and sorted 53 open-source datasets and built a large medical segmentation dataset with 18 modalities, 84 objects, 125 object-modality paired targets, 1050K 2D images, and 6033K masks. We comprehensively analyzed different models and strategies on the so-called COSMOS 1050K dataset. Our findings mainly include the following: 1) SAM showed remarkable performance in some specific objects but was unstable, imperfect, or even totally failed in other situations. 2) SAM with the large ViT-H showed better overall performance than that with the small ViT-B. 3) SAM performed better with manual hints, especially box, than the Everything mode. 4) SAM could help human annotation with high labeling quality and less time. 5) SAM was sensitive to the randomness in the center point and tight box prompts, and may suffer from a serious performance drop. 6) SAM performed better than interactive methods with one or a few points, but will be outpaced as the number of points increases. 7) SAM's performance correlated to different factors, including boundary complexity, intensity differences, etc. 8) Finetuning the SAM on specific medical tasks could improve its average DICE performance by 4.39% and 6.68% for ViT-B and ViT-H, respectively. We hope that this comprehensive report can help researchers explore the potential of SAM applications in MIS, and guide how to appropriately use and develop SAM., Comment: Accepted by Medical Image Analysis. 23 pages, 18 figures, 8 tables

Large pretrained language models are widely used in downstream NLP tasks via task-specific fine-tuning, but such procedures can be costly. Recently, Parameter-Efficient Fine-Tuning (PEFT) methods have achieved strong task performance while updating much fewer parameters than full model fine-tuning (FFT). However, it is non-trivial to make informed design choices on the PEFT configurations, such as their architecture, the number of tunable parameters, and even the layers in which the PEFT modules are inserted. Consequently, it is highly likely that the current, manually designed configurations are suboptimal in terms of their performance-efficiency trade-off. Inspired by advances in neural architecture search, we propose AutoPEFT for automatic PEFT configuration selection: we first design an expressive configuration search space with multiple representative PEFT modules as building blocks. Using multi-objective Bayesian optimisation in a low-cost setup, we then discover a Pareto-optimal set of configurations with strong performance-cost trade-offs across different numbers of parameters that are also highly transferable across different tasks. Empirically, on GLUE and SuperGLUE tasks, we show that AutoPEFT-discovered configurations significantly outperform existing PEFT methods and are on par or better than FFT without incurring substantial training efficiency costs., Comment: Accepted to TACL; pre-MIT Press publication version

Large pre-trained models have revolutionized natural language processing (NLP) research and applications, but high training costs and limited data resources have prevented their benefits from being shared equally amongst speakers of all the world's languages. To address issues of cross-linguistic access to such models and reduce energy consumption for sustainability during large-scale model training, this study proposes an effective and energy-efficient framework called GreenPLM that uses bilingual lexicons to directly "translate" pre-trained language models of one language into another at almost no additional cost. We validate this approach in 18 languages' BERT models and show that this framework is comparable to, if not better than, other heuristics with high training costs. In addition, given lightweight continued pre-training on limited data where available, this framework outperforms the original monolingual language models in six out of seven tested languages with up to 200x less pre-training efforts. Aiming at the Leave No One Behind Principle (LNOB), our approach manages to reduce inequalities between languages and energy consumption greatly. We make our codes and models publicly available here: \url{https://github.com/qcznlp/GreenPLMs}, Comment: Accepted at IJCAI 2023 AI and Social Good Track

Recently, the practical deployment of open-domain dialogue systems has been plagued by the knowledge issue of information deficiency and factual inaccuracy. To this end, we introduce PLATO-K based on two-stage dialogic learning to strengthen internal knowledge memorization and external knowledge exploitation. In the first stage, PLATO-K learns through massive dialogue corpora and memorizes essential knowledge into model parameters. In the second stage, PLATO-K mimics human beings to search for external information and to leverage the knowledge in response generation. Extensive experiments reveal that the knowledge issue is alleviated significantly in PLATO-K with such comprehensive internal and external knowledge enhancement. Compared to the existing state-of-the-art Chinese dialogue model, the overall engagingness of PLATO-K is improved remarkably by 36.2% and 49.2% on chit-chat and knowledge-intensive conversations., Comment: First four authors contributed equally to this work

Light-weight time-of-flight (ToF) depth sensors are small, cheap, low-energy and have been massively deployed on mobile devices for the purposes like autofocus, obstacle detection, etc. However, due to their specific measurements (depth distribution in a region instead of the depth value at a certain pixel) and extremely low resolution, they are insufficient for applications requiring high-fidelity depth such as 3D reconstruction. In this paper, we propose DELTAR, a novel method to empower light-weight ToF sensors with the capability of measuring high resolution and accurate depth by cooperating with a color image. As the core of DELTAR, a feature extractor customized for depth distribution and an attention-based neural architecture is proposed to fuse the information from the color and ToF domain efficiently. To evaluate our system in real-world scenarios, we design a data collection device and propose a new approach to calibrate the RGB camera and ToF sensor. Experiments show that our method produces more accurate depth than existing frameworks designed for depth completion and depth super-resolution and achieves on par performance with a commodity-level RGB-D sensor. Code and data are available at https://zju3dv.github.io/deltar/., Comment: Accepted to ECCV 2022. Project Page: https://zju3dv.github.io/deltar/

Spatial-temporal Map (STMap)-based methods have shown great potential to process high-angle videos for vehicle trajectory reconstruction, which can meet the needs of various data-driven modeling and imitation learning applications. In this paper, we developed Spatial-Temporal Deep Embedding (STDE) model that imposes parity constraints at both pixel and instance levels to generate instance-aware embeddings for vehicle stripe segmentation on STMap. At pixel level, each pixel was encoded with its 8-neighbor pixels at different ranges, and this encoding is subsequently used to guide a neural network to learn the embedding mechanism. At the instance level, a discriminative loss function is designed to pull pixels belonging to the same instance closer and separate the mean value of different instances far apart in the embedding space. The output of the spatial-temporal affinity is then optimized by the mutex-watershed algorithm to obtain final clustering results. Based on segmentation metrics, our model outperformed five other baselines that have been used for STMap processing and shows robustness under the influence of shadows, static noises, and overlapping. The designed model is applied to process all public NGSIM US-101 videos to generate complete vehicle trajectories, indicating a good scalability and adaptability. Last but not least, the strengths of the scanline method with STDE and future directions were discussed. Code, STMap dataset and video trajectory are made publicly available in the online repository. GitHub Link: shorturl.at/jklT0.

Making chatbots world aware in a conversation like a human is a crucial challenge, where the world may contain dynamic knowledge and spatiotemporal state. Several recent advances have tried to link the dialog system to a static knowledge base or search engine, but they do not contain all the world information needed for conversations. In contrast, we propose a new method to improve the dialogue system using spatiotemporal aware dynamic knowledge. We utilize service information as a way for the dialogue system to link the world. The system actively builds a request according to the dialog context and spatiotemporal state to get service information and then generates world aware responses. To implement this method, we collect DuSinc, an open-domain human-human dialogue dataset, where a participant can access the service to get the information needed for dialogue responses. Through automatic and human evaluations, we found that service information significantly improves the consistency, informativeness, factuality, and engagingness of the dialogue system, making it behave more like a human. Compared to the pre-trained models without spatiotemporal aware dynamic knowledge, the overall session-level score was improved by 60.87\%. The collection dataset and methods will be open-sourced., Comment: 13pages, 5 figures

Low-bit width neural networks have been extensively explored for deployment on edge devices to reduce computational resources. Existing approaches have focused on gradient-based optimization in a two-stage train-and-compress setting or as a combined optimization where gradients are quantized during training. Such schemes require high-performance hardware during the training phase and usually store an equivalent number of full-precision weights apart from the quantized weights. In this paper, we explore methods of direct combinatorial optimization in the problem of risk minimization with binary weights, which can be made equivalent to a non-monotone submodular maximization under certain conditions. We employ an approximation algorithm for the cases with single and multilayer neural networks. For linear models, it has $\mathcal{O}(nd)$ time complexity where $n$ is the sample size and $d$ is the data dimension. We show that a combination of greedy coordinate descent and this novel approach can attain competitive accuracy on binary classification tasks.

Information reconciliation (IR), which corrects the errors in the sifted keys, directly determines the secure key rate and the link distance of quantum key distribution (QKD) systems. In this article, we propose an appending information reconciliation (AIR) scheme based on polar codes, which achieves high efficiency and ultra-low failure probability simultaneously, by gradually disclosing the bit values of the polarized channels with high error probability. The experimental results show that the efficiency of the proposed AIR scheme is closer to the Shannon limit, compared with the state-of-the-art implemented polar codes-based IR schemes, with the overall failure probability around 1E-8, especially when performed with smaller block sizes. Moreover, the efficiency of the proposed AIR scheme is 1.046, when the block size is 1 Gb and the quantum bit error rate of 0.02. Therefore, the proposed AIR scheme can further eradicate the performance gap between theory and implementation for QKD systems., Comment: 12 pages, 6 pages, 4 tables

Dialogue State Tracking (DST), a crucial component of task-oriented dialogue (ToD) systems, keeps track of all important information pertaining to dialogue history: filling slots with the most probable values throughout the conversation. Existing methods generally rely on a predefined set of values and struggle to generalise to previously unseen slots in new domains. To overcome these challenges, we propose a domain-agnostic extractive question answering (QA) approach with shared weights across domains. To disentangle the complex domain information in ToDs, we train our DST with a novel domain filtering strategy by excluding out-of-domain question samples. With an independent classifier that predicts the presence of multiple domains given the context, our model tackles DST by extracting spans in active domains. Empirical results demonstrate that our model can efficiently leverage domain-agnostic QA datasets by two-stage fine-tuning while being both domain-scalable and open-vocabulary in DST. It shows strong transferability by achieving zero-shot domain-adaptation results on MultiWOZ 2.1 with an average JGA of 36.7%. It further achieves cross-lingual transfer with state-of-the-art zero-shot results, 66.2% JGA from English to German and 75.7% JGA from English to Italian on WOZ 2.0., Comment: Accepted to Findings of EACL 2023

Accurate segmentation of the blood vessels of the retina is an important step in clinical diagnosis of ophthalmic diseases. Many deep learning frameworks have come up for retinal blood vessels segmentation tasks. However, the complex vascular structure and uncertain pathological features make the blood vessel segmentation still very challenging. A novel U-shaped network named Multi-module Concatenation which is based on Atrous convolution and multi-kernel pooling is put forward to retinal vessels segmentation in this paper. The proposed network structure retains three layers the essential structure of U-Net, in which the atrous convolution combining the multi-kernel pooling blocks are designed to obtain more contextual information. The spatial attention module is concatenated with dense atrous convolution module and multi-kernel pooling module to form a multi-module concatenation. And different dilation rates are selected by cascading to acquire a larger receptive field in atrous convolution. Adequate comparative experiments are conducted on these public retinal datasets: DRIVE, STARE and CHASE_DB1. The results show that the proposed method is effective, especially for microvessels. The code will be put out at https://github.com/Rebeccala/MC-UNet, Comment: 13pages,3957

We propose the family universal $U(1)_X$ models with three right-handed neutrinos by choosing the $U(1)_X$ gauge symmetry as a linear combination of $U(1)_Y\times U(1)_{\chi}$ of $SO(10)$. To be consistent with the quantum gravity effects, we introduced a Dirac fermion $\chi$ as a dark matter candidate, which is odd under the gauged $Z_2$ symmetry after $U(1)_X$ breaking. The isospin violation dark matter with $f_n/f_p = -0.7$ can be realized naturally, and thus the LUX, PANDAX, and XENON1T experimental constraints can be evaded. Moreover, we study the masses and mixings for Higgs and gauge bosons, consider the LHC constraints on the $Z'$ mass, simulate various constraints from dark matter direct and indirect detection experiments, and then present the viable parameter spaces. To study the LHC $Z'$ mass bounds on the generic $U(1)_X$ models, we considered four kinds of scenarios, where scenario II with zero $U(1)_X$ charge for right-handed up-type quarks can relax the LHC $Z'$ mass bound a little bit., Comment: 18 pages, 6 figures

Singular source terms in sub-diffusion equations may lead to the unboundedness of solutions, which will bring a severe reduction of convergence order of existing time-stepping schemes. In this work, we propose two efficient time-stepping schemes for solving sub-diffusion equations with a class of source terms mildly singular in time. One discretization is based on the Gr{\"u}nwald-Letnikov and backward Euler methods. First-order error estimate with respect to time is rigorously established for singular source terms and nonsmooth initial data. The other scheme derived from the second-order backward differentiation formula (BDF) is proved to possess second-order accuracy in time. Further, piecewise linear finite element and lumped mass finite element discretizations in space are applied and analyzed rigorously. Numerical investigations confirm our theoretical results.

Implicit neural rendering techniques have shown promising results for novel view synthesis. However, existing methods usually encode the entire scene as a whole, which is generally not aware of the object identity and limits the ability to the high-level editing tasks such as moving or adding furniture. In this paper, we present a novel neural scene rendering system, which learns an object-compositional neural radiance field and produces realistic rendering with editing capability for a clustered and real-world scene. Specifically, we design a novel two-pathway architecture, in which the scene branch encodes the scene geometry and appearance, and the object branch encodes each standalone object conditioned on learnable object activation codes. To survive the training in heavily cluttered scenes, we propose a scene-guided training strategy to solve the 3D space ambiguity in the occluded regions and learn sharp boundaries for each object. Extensive experiments demonstrate that our system not only achieves competitive performance for static scene novel-view synthesis, but also produces realistic rendering for object-level editing., Comment: Accepted to ICCV 2021. Project Page: https://zju3dv.github.io/object_nerf