Search

Due to the continuous emergence of new data, version updates have become an indispensable requirement for Large Language Models (LLMs). The training paradigms for version updates of LLMs include pre-training from scratch (PTFS) and continual pre-training (CPT). Preliminary experiments demonstrate that PTFS achieves better pre-training performance, while CPT has lower training cost. Moreover, their performance and training cost gaps widen progressively with version updates. To investigate the underlying reasons for this phenomenon, we analyze the effect of learning rate adjustments during the two stages of CPT: preparing an initialization checkpoint and continual pre-training based on this checkpoint. We find that a large learning rate in the first stage and a complete learning rate decay process in the second stage are crucial for version updates of LLMs. Hence, we propose a learning rate path switching training paradigm. Our paradigm comprises one main path, where we pre-train a LLM with the maximal learning rate, and multiple branching paths, each of which corresponds to an update of the LLM with newly-added training data. Extensive experiments demonstrate the effectiveness and generalization of our paradigm. Particularly, when training four versions of LLMs, our paradigm reduces the total training cost to 58% compared to PTFS, while maintaining comparable pre-training performance., Comment: EMNLP 2024 (main,long paper)

Multimodal task specification is essential for enhanced robotic performance, where \textit{Cross-modality Alignment} enables the robot to holistically understand complex task instructions. Directly annotating multimodal instructions for model training proves impractical, due to the sparsity of paired multimodal data. In this study, we demonstrate that by leveraging unimodal instructions abundant in real data, we can effectively teach robots to learn multimodal task specifications. First, we endow the robot with strong \textit{Cross-modality Alignment} capabilities, by pretraining a robotic multimodal encoder using extensive out-of-domain data. Then, we employ two Collapse and Corrupt operations to further bridge the remaining modality gap in the learned multimodal representation. This approach projects different modalities of identical task goal as interchangeable representations, thus enabling accurate robotic operations within a well-aligned multimodal latent space. Evaluation across more than 130 tasks and 4000 evaluations on both simulated LIBERO benchmark and real robot platforms showcases the superior capabilities of our proposed framework, demonstrating significant advantage in overcoming data constraints in robotic learning. Website: zh1hao.wang/Robo_MUTUAL, Comment: preprint

In the paper, we show some properties of (reduced) stated SL($n$)-skein algebras related to their centers for essentially bordered pb surfaces, especially their centers, finitely generation over their centers, and their PI-degrees. The proofs are based on the quantum trace maps, embeddings of (reduced) stated SL($n$)-skein algebras into quantum tori appearing in higher Teichm\"uller theory. Thanks to the Unicity theorem in [BG02, FKBL19], we can understand the representation theory of (reduced) stated SL($n$)-skein algebras. Moreover, the applications are beyond low-dimensional topology. For example, we can access to the representation theory of unrestricted quantum moduli algebras, and that of quantum higher cluster algebras potentially., Comment: 70 pages, 12 figures

Microservice-based systems often suffer from reliability issues due to their intricate interactions and expanding scale. With the rapid growth of observability techniques, various methods have been proposed to achieve failure diagnosis, including root cause localization and failure type identification, by leveraging diverse monitoring data such as logs, metrics, or traces. However, traditional failure diagnosis methods that use single-modal data can hardly cover all failure scenarios due to the restricted information. Several failure diagnosis methods have been recently proposed to integrate multimodal data based on deep learning. These methods, however, tend to combine modalities indiscriminately and treat them equally in failure diagnosis, ignoring the relationship between specific modalities and different diagnostic tasks. This oversight hinders the effective utilization of the unique advantages offered by each modality. To address the limitation, we propose \textit{TVDiag}, a multimodal failure diagnosis framework for locating culprit microservice instances and identifying their failure types (e.g., Net-packets Corruption) in microservice-based systems. \textit{TVDiag} employs task-oriented learning to enhance the potential advantages of each modality and establishes cross-modal associations based on contrastive learning to extract view-invariant failure information. Furthermore, we develop a graph-level data augmentation strategy that randomly inactivates the observability of some normal microservice instances during training to mitigate the shortage of training data. Experimental results show that \textit{TVDiag} outperforms state-of-the-art methods in multimodal failure diagnosis, achieving at least a 55.94\% higher $HR@1$ accuracy and over a 4.08\% increase in F1-score across two datasets., Comment: 32 pages

The ${\rm SL}_3$-skein algebra $\mathscr{S}_{\bar{q}}(\mathfrak{S})$ of a punctured oriented surface $\mathfrak{S}$ is a quantum deformation of the coordinate algebra of the ${\rm SL}_3$-character variety of $\mathfrak{S}$. When $\bar{q}$ is a root of unity, we prove the Unicity Theorem for representations of $\mathscr{S}_{\bar{q}}(\mathfrak{S})$, in particular the existence and uniqueness of a generic irreducible representation. Furthermore, we show that the center of $\mathscr{S}_{\bar{q}}(\frak{S})$ is generated by the peripheral skeins around punctures and the central elements contained in the image of the Frobenius homomorphism for $\mathscr{S}_{\bar{q}}(\frak{S})$, a surface generalization of Frobenius homomorphisms of quantum groups related to ${\rm SL}_3$. We compute the rank of $\mathscr{S}_{\bar{q}}(\mathfrak{S})$ over its center, hence the dimension of the generic irreducible representation., Comment: v2: 56 pages, reorganized, streamlined, added some details

The intrinsic symmetries of physical systems have been employed to reduce the number of degrees of freedom of systems, thereby simplifying computations. In this work, we investigate the properties of $\mathcal{M}SU(2^N)$, $\mathcal{M}$-invariant subspaces of the special unitary Lie group $SU(2^N)$ acting on $N$ qubits, for some $\mathcal{M}\subseteq M_{2^N}(\mathbb{C})$. We demonstrate that for certain choices of $\mathcal{M}$, the subset $\mathcal{M}SU(2^N)$ inherits many topological and group properties from $SU(2^N)$. We then present a combinatorial method for computing the dimension of such subspaces when $\mathcal{M}$ is a representation of a permutation group acting on qubits $(GSU(2^N))$, or a Hamiltonian $(H^{(N)}SU(2^N))$. The Kronecker product of $\mathfrak{su}(2)$ matrices is employed to construct the Lie algebras associated with different permutation-invariant groups $GSU(2^N)$. Numerical results on the number of dimensions support the the developed theory., Comment: 27 pages, 6 figures, 1 table

Token repetition is a typical form of multi-modal problem in fully non-autoregressive translation (NAT). In this work, we revisit the multi-modal problem in recently proposed NAT models. Our study reveals that these advanced models have introduced other types of information redundancy errors, which cannot be measured by the conventional metric - the continuous repetition ratio. By manually annotating the NAT outputs, we identify two types of information redundancy errors that correspond well to lexical and reordering multi-modality problems. Since human annotation is time-consuming and labor-intensive, we propose automatic metrics to evaluate the two types of redundant errors. Our metrics allow future studies to evaluate new methods and gain a more comprehensive understanding of their effectiveness., Comment: 10 pages, 10 tables

Foundation models, i.e., very large deep learning models, have demonstrated impressive performances in various language and vision tasks that are otherwise difficult to reach using smaller-size models. The major success of GPT-type of language models is particularly exciting and raises expectations on the potential of foundation models in other domains including satellite remote sensing. In this context, great efforts have been made to build foundation models to test their capabilities in broader applications, and examples include Prithvi by NASA-IBM, Segment-Anything-Model, ViT, etc. This leads to an important question: Are foundation models always a suitable choice for different remote sensing tasks, and when or when not? This work aims to enhance the understanding of the status and suitability of foundation models for pixel-level classification using multispectral imagery at moderate resolution, through comparisons with traditional machine learning (ML) and regular-size deep learning models. Interestingly, the results reveal that in many scenarios traditional ML models still have similar or better performance compared to foundation models, especially for tasks where texture is less useful for classification. On the other hand, deep learning models did show more promising results for tasks where labels partially depend on texture (e.g., burn scar), while the difference in performance between foundation models and deep learning models is not obvious. The results conform with our analysis: The suitability of foundation models depend on the alignment between the self-supervised learning tasks and the real downstream tasks, and the typical masked autoencoder paradigm is not necessarily suitable for many remote sensing problems.

Human motion prediction is consisting in forecasting future body poses from historically observed sequences. It is a longstanding challenge due to motion's complex dynamics and uncertainty. Existing methods focus on building up complicated neural networks to model the motion dynamics. The predicted results are required to be strictly similar to the training samples with L2 loss in current training pipeline. However, little attention has been paid to the uncertainty property which is crucial to the prediction task. We argue that the recorded motion in training data could be an observation of possible future, rather than a predetermined result. In addition, existing works calculate the predicted error on each future frame equally during training, while recent work indicated that different frames could play different roles. In this work, a novel computationally efficient encoder-decoder model with uncertainty consideration is proposed, which could learn proper characteristics for future frames by a dynamic function. Experimental results on benchmark datasets demonstrate that our uncertainty consideration approach has obvious advantages both in quantity and quality. Moreover, the proposed method could produce motion sequences with much better quality that avoids the intractable shaking artefacts. We believe our work could provide a novel perspective to consider the uncertainty quality for the general motion prediction task and encourage the studies in this field. The code will be available in https://github.com/Motionpre/Adaptive-Salient-Loss-SAGGB., Comment: Accepted by AAAI2024

Metal thin films on soft polymers provide a unique opportunity for resistance-based strain sensors. A mechanical mismatch between the conductive film and the flexible substrate causes cracks to open and close, changing the electrical resistance as a function of strain. However, the very randomness of the formation, shape, length, orientation, and distance between adjacent cracks limits the sensing range as well as repeatability. Herein, we present a breakthrough: the Zerogap Strain Sensor, whereby lithography eliminates randomness and violent tearing process inherent in conventional crack sensors and allows for short periodicity between gaps with gentle sidewall contacts, critical in high strain sensing enabling operation over an unprecedently wide range. Our sensor achieves a gauge factor of over 15,000 at {\epsilon}ext=18%, the highest known value. With the uniform gaps of three-to-ten thousand nanometer widths characterized by periodicity and strain, this approach has far reaching implications for future strain sensors whose range is limited only by that of the flexible substrate, with non-violent operations that always remain below the tensile limit of the metal.

We show that quantum circuits restricted by a symmetry inherit the properties of the whole special unitary group $SU(2^n)$, in particular composition, algebraic and topological closedness and connectedness. It extends prior work on symmetric states to the operators and shows that the operator space follows the same structure as the state space. The well-behavedness is independent of the symmetry requirement imposed on the subgroup. We provide an example of a permutation invariance across all qubits., Comment: 10 pages, 2 figures

Fairness-awareness has emerged as an essential building block for the responsible use of artificial intelligence in real applications. In many cases, inequity in performance is due to the change in distribution over different regions. While techniques have been developed to improve the transferability of fairness, a solution to the problem is not always feasible with no samples from the new regions, which is a bottleneck for pure data-driven attempts. Fortunately, physics-based mechanistic models have been studied for many problems with major social impacts. We propose SimFair, a physics-guided fairness-aware learning framework, which bridges the data limitation by integrating physical-rule-based simulation and inverse modeling into the training design. Using temperature prediction as an example, we demonstrate the effectiveness of the proposed SimFair in fairness preservation., Comment: Accepted to AAAI 2024 (preprint)

Let $(M,\mathcal{N})$ be a marked 3-manifold. We use $S_n(M,\mathcal{N},v)$ to denote the stated $SL_n$-skein module of $(M,\mathcal{N})$ where $v$ is a nonzero complex number. We establish a surjective algebra homomorphism from $S_n(M,\mathcal{N},1)$ to the coordinate ring of some algebraic set, and prove its kernel consists of all nilpotents. We prove the universal representation algebra of $\pi_1(M)$ is isomorphic to $S_n(M,\mathcal{N},1)$ when $M$ is connected and $\mathcal{N}$ has only one component. Furthermore, we show $S_n(M,\mathcal{N}',1)$ is isomorphic to $S_n(M,\mathcal{N},1)\otimes O(SL_n)$ as algebras, where $(M,\mathcal{N})$ is a connected marked 3-manifold with $\mathcal{N}\neq\emptyset$, and $\mathcal{N}'$ is obtained from $\mathcal{N}$ by adding one extra marking., Comment: We split our submitted paper "arXiv:2307.10288" into two papers. The current one contains sections 3, 4, 5, and 9 of the original version, and has 35 pages. We will use the other one to replace the original copy

We study a generalized Witten's finiteness conjecture for the skein modules of oriented compact 3-manifolds with boundary. We formulate an equivalent version of the generalized finiteness conjecture using handlebodies and 2-handles, and prove the conjecture for some classes with the handlebodies of genus 2 and 3 using the equivalent version., Comment: 16 pages, 10 figures

For any marked three manifold $(M,\mathcal N)$ and any quantum parameter $q^{\frac{1}{2}}$ (a nonzero complex number), we use $\mathscr{S}_{q^{1/2}}(M,\mathcal{N})$ to denote the stated skein module of $(M,\mathcal{N})$. When $q^{\frac{1}{2}}$ is a root of unity of odd order, the commutative algebra $\mathscr{S}_1(M,\mathcal{N})$ acts on $\mathscr{S}_{q^{1/2}}(M,\mathcal{N})$. For any maximal ideal $\rho$ of $\mathscr{S}_1(M,\mathcal{N})$, define $\mathscr{S}_{q^{1/2}}(M,\mathcal{N})_{\rho} = \mathscr{S}_{q^{1/2}}(M,\mathcal{N})\otimes _{\mathscr{S}_1(M,\mathcal{N})} (\mathscr{S}_1(M,\mathcal{N})/\rho)$. We prove the splitting map for $\mathscr{S}_{q^{1/2}}(M,\mathcal{N})$ respects the $\mathscr{S}_1(M,\mathcal{N})$-module structure, so it reduces to the splitting map for $\mathscr{S}_{q^{1/2}}(M,\mathcal{N})_{\rho}$. We prove the splitting map for $\mathscr{S}_{q^{1/2}}(M,\mathcal{N})_{\rho}$ is injective if there exists at least one component of $\mathcal{N}$ such that this component and the boundary of the splitting disk belong to the same component of $\partial M$. We also prove the representation-reduced stated skein module of the marked handlebody is an irreducible Azumaya representation of the stated skein algebra of its boundary. Let $M$ be an oriented connected closed three manifold. For any positive integer $k$, we use $M_{k}$ to denote the marked three manifold obtained from $M$ by removing $k$ open three dimensional balls and adding one marking to each newly created sphere boundary component. We prove $\text{dim}_{\mathbb{C}}\mathscr{S}_{q^{1/2}}( M_{k})_{\rho} = 1$ for any maximal ideal $\rho$ of $\mathscr{S}_1(M_k)$., Comment: title change, the update for published version, accepted by Journal of Algebra

When the quantum parameter $q^{\frac{1}{2}}$ is a root of unity of odd order and the punctured bordered surface has nonempty boundary, we prove the fraction ring of the stated skein algebra (that is the localization over all nonzero elements) is a symmetric Frobenius algebra over both the field of fractions of the image of the Frobenius map and the field of fractions of the center of the stated skein algebra. We also calculate Traces of the fraction ring of the stated skein algebra over these two fields., Comment: 21 pages

Existing satellite remote sensing change detection (CD) methods often crop original large-scale bi-temporal image pairs into small patch pairs and then use pixel-level CD methods to fairly process all the patch pairs. However, due to the sparsity of change in large-scale satellite remote sensing images, existing pixel-level CD methods suffer from a waste of computational cost and memory resources on lots of unchanged areas, which reduces the processing efficiency of on-board platform with extremely limited computation and memory resources. To address this issue, we propose a lightweight patch-level CD network (LPCDNet) to rapidly remove lots of unchanged patch pairs in large-scale bi-temporal image pairs. This is helpful to accelerate the subsequent pixel-level CD processing stage and reduce its memory costs. In our LPCDNet, a sensitivity-guided channel pruning method is proposed to remove unimportant channels and construct the lightweight backbone network on basis of ResNet18 network. Then, the multi-layer feature compression (MLFC) module is designed to compress and fuse the multi-level feature information of bi-temporal image patch. The output of MLFC module is fed into the fully-connected decision network to generate the predicted binary label. Finally, a weighted cross-entropy loss is utilized in the training process of network to tackle the change/unchange class imbalance problem. Experiments on two CD datasets demonstrate that our LPCDNet achieves more than 1000 frames per second on an edge computation platform, i.e., NVIDIA Jetson AGX Orin, which is more than 3 times that of the existing methods without noticeable CD performance loss. In addition, our method reduces more than 60% memory costs of the subsequent pixel-level CD processing stage.

This paper introduces a novel class of fair and interpolatory curves called $p\kappa$-curves. These curves are comprised of smoothly stitched B\'ezier curve segments, where the curvature distribution of each segment is made to closely resemble a parabola, resulting in an aesthetically pleasing shape. Moreover, each segment passes through an interpolated point at a parameter where the parabola has an extremum, encouraging the alignment of interpolated points with curvature extrema. To achieve these properties, we tailor an energy function that guides the optimization process to obtain the desired curve characteristics. Additionally, we develop an efficient algorithm and an initialization method, enabling interactive modeling of the $p\kappa$-curves without the need for global optimization. We provide various examples and comparisons with existing state-of-the-art methods to demonstrate the curve modeling capabilities and visually pleasing appearance of $p\kappa$-curves.

When the quantum parameter $q^{1/2}$ is a root of unity of odd order. The stated skein module $S_{q^{1/2}}(M,\mathcal{N})$ has an $S_{1}(M,\mathcal{N})$-module structure, where $(M,\mathcal{N})$ is a marked three manifold. We prove $S_{q^{1/2}}(M,\mathcal{N})$ is a finitely generated $S_{1}(M,\mathcal{N})$-module when $M$ is compact, which furthermore indicates the reduced stated skein module for the compact marked three manifold is finite dimensional. We also give an upper bound for the dimension of $S_{q^{1/2}}(M,\mathcal{N})$ over $S_{1}(M,\mathcal{N})$ when $M$ is compact. For a pb surface $\Sigma$, we use $S_{q^{1/2}}(\Sigma)^{(N)}$ to denote the image of the Frobenius map when $q^{1/2}$ is a root of unity of odd order $N$. Then $S_{q^{1/2}}(\Sigma)^{(N)}$ lives in the center of the stated skein algebra $S_{q^{1/2}}(\Sigma)$. Let $\widetilde{S_{q^{1/2}}(\Sigma)^{(N)}}$ be the field of fractions of $S_{q^{1/2}}(\Sigma)^{(N)}$, and $\widetilde{S_{q^{1/2}}(\Sigma)}$ be $S_{q^{1/2}}(\Sigma)\otimes_{S_{q^{1/2}}(\Sigma)^{(N)}} \widetilde{S_{q^{1/2}}(\Sigma)^{(N)}}$. Then we show the dimension of $\widetilde{S_{q^{1/2}}(\Sigma)}$ over $\widetilde{S_{q^{1/2}}(\Sigma)^{(N)}}$ is $N^{3r(\Sigma)}$ where $r(\Sigma)$ equals to the number of boundary components of $\Sigma$ minus the Euler characteristic of $\Sigma$., Comment: 31 pages

Multi-object tracking (MOT) has important applications in monitoring, logistics, and other fields. This paper develops a real-time multi-object tracking and prediction system in rugged environments. A 3D object detection algorithm based on Lidar-camera fusion is designed to detect the target objects. Based on the Hungarian algorithm, this paper designs a 3D multi-object tracking algorithm with an adaptive threshold to realize the stable matching and tracking of the objects. We combine Memory Augmented Neural Networks (MANN) and Kalman filter to achieve 3D trajectory prediction on rugged terrains. Besides, we realize a new dynamic SLAM by using the results of multi-object tracking to remove dynamic points for better SLAM performance and static map. To verify the effectiveness of the proposed multi-object tracking and prediction system, several simulations and physical experiments are conducted. The results show that the proposed system can track dynamic objects and provide future trajectory and a more clean static map in real-time., Comment: 6 Pages, 8 figures, submitted to ROBIO2023

We mainly focus on Classical limit, Splitting map, and Frobenius homomorphism for stated $SL(n)$-skein modules, and Unicity Theorem for stated $SL(n)$-skein algebras. Let $(M,N)$ be a marked three manifold. We use $S_n(M,N,v)$ to denote the stated $SL(n)$-skein module of $(M,N)$ where $v$ is a nonzero complex number. We build a surjective algebra homomorphism from $S_n(M,N,1)$ to the coordinate ring of some algebraic set, and prove it's Kernal consists of all nilpotents. We prove the universal representation algebra of $\pi_1(M)$ is isomorphic to $S_n(M,N,1)$ when $N$ has only one component and $M$ is connected. Furthermore we show $S_n(M,N^{'},1)$ is isomorphic to $S_n(M,N,1)\otimes O(SLn)$, where $N\neq \emptyset$, $M$ is connected, and $N^{'}$ is obtained from $N$ by adding one extra marking. We also prove the splitting map is injective for any marked three manifold when $v=1$, and show that the splitting map is injective (for general $v$) if there exists at least one component of $N$ such that this component and the boundary of the splitting disk belong to the same component of $\partial M$. We also establish the Frobenius homomorphism for $SL(n)$, which is map from $S_n(M,N,1)$ to $S_n(M,N,v)$ when $v$ is a primitive $m$-th root of unity with $m$ being coprime with $2n$ and every component of $M$ contains at least one marking. We also show the commutativity between Frobenius homomorphism and splitting map. When $(M,N)$ is the thickening of an essentially bordered pb surface, we prove the Frobenius homomorphism is injective and it's image lives in the center. We prove the stated $SL(n)$-skein algebra $S_n(\Sigma,v)$ is affine almost Azumaya when $\Sigma$ is an essentially bordered pb surface and $v$ is a primitive $m$-th root of unity with $m$ being coprime with $2n$, which implies the Unicity Theorem for $S_n(\Sigma,v)$., Comment: 66 pages. We rewrote the construction for Frobenious map and a module structure for stated $SL(n)$-skein module in section 7

Recently, the text-to-table generation task has attracted increasing attention due to its wide applications. In this aspect, the dominant model formalizes this task as a sequence-to-sequence generation task and serializes each table into a token sequence during training by concatenating all rows in a top-down order. However, it suffers from two serious defects: 1) the predefined order introduces a wrong bias during training, which highly penalizes shifts in the order between rows; 2) the error propagation problem becomes serious when the model outputs a long token sequence. In this paper, we first conduct a preliminary study to demonstrate the generation of most rows is order-insensitive. Furthermore, we propose a novel sequence-to-sequence&set text-to-table generation model. Specifically, in addition to a text encoder encoding the input text, our model is equipped with a table header generator to first output a table header, i.e., the first row of the table, in the manner of sequence generation. Then we use a table body generator with learnable row embeddings and column embeddings to generate a set of table body rows in parallel. Particularly, to deal with the issue that there is no correspondence between each generated table body row and target during training, we propose a target assignment strategy based on the bipartite matching between the first cells of generated table body rows and targets. Experiment results show that our model significantly surpasses the baselines, achieving state-of-the-art performance on commonly-used datasets.

In real-world systems, scaling has been critical for improving the translation quality in autoregressive translation (AT), which however has not been well studied for non-autoregressive translation (NAT). In this work, we bridge the gap by systematically studying the impact of scaling on NAT behaviors. Extensive experiments on six WMT benchmarks over two advanced NAT models show that scaling can alleviate the commonly-cited weaknesses of NAT models, resulting in better translation performance. To reduce the side-effect of scaling on decoding speed, we empirically investigate the impact of NAT encoder and decoder on the translation performance. Experimental results on the large-scale WMT20 En-De show that the asymmetric architecture (e.g. bigger encoder and smaller decoder) can achieve comparable performance with the scaling model, while maintaining the superiority of decoding speed with standard NAT models. To this end, we establish a new benchmark by validating scaled NAT models on the scaled dataset, which can be regarded as a strong baseline for future works. We release code and system outputs at https://github.com/DeepLearnXMU/Scaling4NAT., Comment: 13 pages, Findings of ACL 2023

This paper presents the application of triangle configuration B-splines (TCB-splines) for representing and analyzing the Kirchhoff-Love shell in the context of isogeometric analysis (IGA). The Kirchhoff-Love shell formulation requires global $C^1$-continuous basis functions. The nonuniform rational B-spline (NURBS)-based IGA has been extensively used for developing Kirchhoff-Love shell elements. However, shells with complex geometries inevitably need multiple patches and trimming techniques, where stitching patches with high continuity is a challenge. On the other hand, due to their unstructured nature, TCB-splines can accommodate general polygonal domains, have local refinement, and are flexible to model complex geometries with $C^1$ continuity, which naturally fit into the Kirchhoff-Love shell formulation with complex geometries. Therefore, we propose to use TCB-splines as basis functions for geometric representation and solution approximation. We apply our method to both linear and nonlinear benchmark shell problems, where the accuracy and robustness are validated. The applicability of the proposed approach to shell analysis is further exemplified by performing geometrically nonlinear Kirchhoff-Love shell simulations of a pipe junction and a front bumper represented by a single patch of TCB-splines.

Label noise is ubiquitous in various machine learning scenarios such as self-labeling with model predictions and erroneous data annotation. Many existing approaches are based on heuristics such as sample losses, which might not be flexible enough to achieve optimal solutions. Meta learning based methods address this issue by learning a data selection function, but can be hard to optimize. In light of these pros and cons, we propose Selection-Enhanced Noisy label Training (SENT) that does not rely on meta learning while having the flexibility of being data-driven. SENT transfers the noise distribution to a clean set and trains a model to distinguish noisy labels from clean ones using model-based features. Empirically, on a wide range of tasks including text classification and speech recognition, SENT improves performance over strong baselines under the settings of self-training and label corruption.

The volume conjecture relates the quantum invariant and the hyperbolic geometry. Bonahon-Wong-Yang introduced a new version of the volume conjecture by using the intertwiners between two isomorphic irreducible representations of the skein algebra. The intertwiners are built from surface diffeomorphisms; they formulated the volume conjecture when diffeomorphisms are pseudo-Anosov. In this paper, we explicitly calculate all the intertwiners for the closed torus using an algebraic embedding from the skein algebra of the closed torus to a quantum torus, and show the limit superior related to the trace of these intertwiners is zero. Moreover, we consider the periodic diffeomorphisms for surfaces with negative Euler characteristic, and conjecture the corresponding limit is zero because the simplicial volume of the mapping tori for periodic diffeomorphisms is zero. For the once punctured torus, we make precise calculations for intertwiners and prove our conjecture., Comment: 35 pages, the update for the accepted version, to appear in Algebraic & Geometric Topology

The visual simultaneous localization and mapping(vSLAM) is widely used in GPS-denied and open field environments for ground and surface robots. However, due to the frequent perception failures derived from lacking visual texture or the {swing} of robot view direction on rough terrains, the accuracy and robustness of vSLAM are still to be enhanced. The study develops a novel view planning approach of actively perceiving areas with maximal information to address the mentioned problem; a gimbal camera is used as the main sensor. Firstly, a map representation based on feature distribution-weighted Fisher information is proposed to completely and effectively represent environmental information richness. With the map representation, a continuous environmental information model is further established to convert the discrete information space into a continuous one for numerical optimization in real-time. Subsequently, the receding horizon optimization is utilized to obtain the optimal informative viewpoints with simultaneously considering the robotic perception, exploration and motion cost based on the continuous environmental model. Finally, several simulations and outdoor experiments are performed to verify the improvement of localization robustness and accuracy by the proposed approach., Comment: 11 pages, 14 figures, in IEEE/ASME Transactions on Mechatronics

Explaining deep learning models is of vital importance for understanding artificial intelligence systems, improving safety, and evaluating fairness. To better understand and control the CNN model, many methods for transparency-interpretability have been proposed. However, most of these works are less intuitive for human understanding and have insufficient human control over the CNN model. We propose a novel method, Transparent and Controllable Network Learning (TCNL), to overcome such challenges. Towards the goal of improving transparency-interpretability, in TCNL, we define some concepts for specific classification tasks through scientific human-intuition study and incorporate concept information into the CNN model. In TCNL, the shallow feature extractor gets preliminary features first. Then several concept feature extractors are built right after the shallow feature extractor to learn high-dimensional concept representations. The concept feature extractor is encouraged to encode information related to the predefined concepts. We also build the concept mapper to visualize features extracted by the concept extractor in a human-intuitive way. TCNL provides a generalizable approach to transparency-interpretability. Researchers can define concepts corresponding to certain classification tasks and encourage the model to encode specific concept information, which to a certain extent improves transparency-interpretability and the controllability of the CNN model. The datasets (with concept sets) for our experiments will also be released (https://github.com/bupt-ai-cz/TCNL).

Efficient video recognition is a hot-spot research topic with the explosive growth of multimedia data on the Internet and mobile devices. Most existing methods select the salient frames without awareness of the class-specific saliency scores, which neglect the implicit association between the saliency of frames and its belonging category. To alleviate this issue, we devise a novel Temporal Saliency Query (TSQ) mechanism, which introduces class-specific information to provide fine-grained cues for saliency measurement. Specifically, we model the class-specific saliency measuring process as a query-response task. For each category, the common pattern of it is employed as a query and the most salient frames are responded to it. Then, the calculated similarities are adopted as the frame saliency scores. To achieve it, we propose a Temporal Saliency Query Network (TSQNet) that includes two instantiations of the TSQ mechanism based on visual appearance similarities and textual event-object relations. Afterward, cross-modality interactions are imposed to promote the information exchange between them. Finally, we use the class-specific saliencies of the most confident categories generated by two modalities to perform the selection of salient frames. Extensive experiments demonstrate the effectiveness of our method by achieving state-of-the-art results on ActivityNet, FCVID and Mini-Kinetics datasets. Our project page is at https://lawrencexia2008.github.io/projects/tsqnet ., Comment: Accepted by ECCV 2022

The machine learning (ML) and deep learning (DL) techniques are widely recognized to be powerful tools for virtual drug screening. The recently reported ML- or DL-based scoring functions have shown exciting performance in predicting protein-ligand binding affinities with fruitful application prospects. However, the differentiation between highly similar ligand conformations, including the native binding pose (the global energy minimum state), remains challenging which could greatly enhance the docking. In this work, we propose a fully differentiable framework for ligand pose optimization based on a hybrid scoring function (SF) combined with a multi-layer perceptron (DeepRMSD) and the traditional AutoDock Vina SF. The DeepRMSD+Vina, which combines (1) the root mean square deviation (RMSD) of the docking pose with respect to the native pose and (2) the AutoDock Vina score, is fully differentiable thus is capable of optimizing the ligand binding pose to the energy-lowest conformation. Evaluated by the CASF-2016 docking power dataset, the DeepRMSD+Vina reaches a success rate of 95.4%, which is by far the best reported SF to date. Based on this SF, an end-to-end ligand pose optimization framework was implemented to improve the docking pose quality. We demonstrated that this method significantly improves the docking success rate (by 15%) in redocking and crossdocking tasks, revealing the high potentialities of this framework in drug design and discovery.

Context modeling plays a significant role in building multi-turn dialogue systems. In order to make full use of context information, systems can use Incomplete Utterance Rewriting(IUR) methods to simplify the multi-turn dialogue into single-turn by merging current utterance and context information into a self-contained utterance. However, previous approaches ignore the intent consistency between the original query and rewritten query. The detection of omitted or coreferred locations in the original query can be further improved. In this paper, we introduce contrastive learning and multi-task learning to jointly model the problem. Our method benefits from carefully designed self-supervised objectives, which act as auxiliary tasks to capture semantics at both sentence-level and token-level. The experiments show that our proposed model achieves state-of-the-art performance on several public datasets.

At present, the existing gait recognition systems are focusing on developing methods to extract robust gait feature from silhouette images and they indeed achieved great success. However, gait can be sensitive to appearance features such as clothing and carried items. Compared with appearance-based method, model-based gait recognition is promising due to the robustness against these variations. In recent years, with the development of human pose estimation, the difficulty of model-based gait recognition methods has been mitigated. In this paper, to resist the increase of subjects and views variation, local features are built and a siamese network is proposed to maximize the distance of samples from the same subject. We leverage recent advances in action recognition to embed human pose sequence to a vector and introduce Spatial-Temporal Graph Convolution Blocks (STGCB) which has been commonly used in action recognition for gait recognition. Experiments on the very large population dataset named OUMVLP-Pose and the popular dataset, CASIA-B, show that our method archives some state-of-the-art (SOTA) performances in model-based gait recognition. The code and models of our method are available at https://github.com/timelessnaive/Gait-for-Large-Dataset after being accepted.

Image super-resolution (SR) methods can generate remote sensing images with high spatial resolution without increasing the cost, thereby providing a feasible way to acquire high-resolution remote sensing images, which are difficult to obtain due to the high cost of acquisition equipment and complex weather. Clearly, image super-resolution is a severe ill-posed problem. Fortunately, with the development of deep learning, the powerful fitting ability of deep neural networks has solved this problem to some extent. In this paper, we propose a network based on the generative adversarial network (GAN) to generate high resolution remote sensing images, named the multi-attention generative adversarial network (MA-GAN). We first designed a GAN-based framework for the image SR task. The core to accomplishing the SR task is the image generator with post-upsampling that we designed. The main body of the generator contains two blocks; one is the pyramidal convolution in the residual-dense block (PCRDB), and the other is the attention-based upsample (AUP) block. The attentioned pyramidal convolution (AttPConv) in the PCRDB block is a module that combines multi-scale convolution and channel attention to automatically learn and adjust the scaling of the residuals for better results. The AUP block is a module that combines pixel attention (PA) to perform arbitrary multiples of upsampling. These two blocks work together to help generate better quality images. For the loss function, we design a loss function based on pixel loss and introduce both adversarial loss and feature loss to guide the generator learning. We have compared our method with several state-of-the-art methods on a remote sensing scene image dataset, and the experimental results consistently demonstrate the effectiveness of the proposed MA-GAN.

Offline Chinese handwriting text recognition is a long-standing research topic in the field of pattern recognition. In previous studies, text detection and recognition are separated, which leads to the fact that text recognition is highly dependent on the detection results. In this paper, we propose a robust end-to-end Chinese text page spotter framework. It unifies text detection and text recognition with text kernel that integrates global text feature information to optimize the recognition from multiple scales, which reduces the dependence of detection and improves the robustness of the system. Our method achieves state-of-the-art results on the CASIA-HWDB2.0-2.2 dataset and ICDAR-2013 competition dataset. Without any language model, the correct rates are 99.12% and 94.27% for line-level recognition, and 99.03% and 94.20% for page-level recognition, respectively.

Localization on 3D data is a challenging task for unmanned vehicles, especially in long-term dynamic urban scenarios. Due to the generality and long-term stability, the pole-like objects are very suitable as landmarks for unmanned vehicle localization in time-varing scenarios. In this paper, a long-term LiDAR-only localization algorithm based on semantic cluster map is proposed. At first, the Convolutional Neural Network(CNN) is used to infer the semantics of LiDAR point clouds. Combined with the point cloud segmentation, the long-term static objects pole/trunk in the scene are extracted and registered into a semantic cluster map. When the unmanned vehicle re-enters the environment again, the relocalization is completed by matching the clusters of the local map with the clusters of the global map. Furthermore, the continuous matching between the local and global maps stably outputs the global pose at 2Hz to correct the drift of the 3D LiDAR odometry. The proposed approach realizes localization in the long-term scenarios without maintaining the high-precision point cloud map. The experimental results on our campus dataset demonstrate that the proposed approach performs better in localization accuracy compared with the current state-of-the-art methods. The source of this paper is available at: http://www.github.com/HITSZ-NRSL/long-term-localization.

In this work, we introduce {\em topological representations of a quiver} as a system consisting of topological spaces and its relationships determined by the quiver. Such a setting gives a natural connection between topological representations of a quiver and diagrams of topological spaces. First, we investigate the relation between the category of topological representations and that of linear representations of a quiver via $P(\Gamma)$-$\mathcal{TOP}^o$ and $k\Gamma$-Mod, concerning (positively) graded or vertex (positively) graded modules. Second, we discuss the homological theory of topological representations of quivers via $\Gamma$-limit $Lim^{\Gamma}$ and using it, define the homology groups of topological representations of quivers via $H_n$. It is found that some properties of a quiver can be read from homology groups. Third, we investigate the homotopy theory of topological representations of quivers. We define the homotopy equivalence between two morphisms in $\textbf{Top}\mathrm{-}\textbf{Rep}\Gamma$ and show that the parallel Homotopy Axiom also holds for top-representations based on the homotopy equivalence. Last, we mainly obtain the functor $At^{\Gamma}$ from $\textbf{Top}\mathrm{-}\textbf{Rep}\Gamma$ to $\textbf{Top}$ and show that $At^{\Gamma}$ preserves homotopy equivalence between morphisms. The relationship is established between the homotopy groups of a top-representation $(T,f)$ and the homotopy groups of $At^{\Gamma}(T,f)$., Comment: 32 pages

Meta-learning methods have been extensively studied and applied in computer vision, especially for few-shot classification tasks. The key idea of meta-learning for few-shot classification is to mimic the few-shot situations faced at test time by randomly sampling classes in meta-training data to construct few-shot tasks for episodic training. While a rich line of work focuses solely on how to extract meta-knowledge across tasks, we exploit the complementary problem on how to generate informative tasks. We argue that the randomly sampled tasks could be sub-optimal and uninformative (e.g., the task of classifying "dog" from "laptop" is often trivial) to the meta-learner. In this paper, we propose an adaptive task sampling method to improve the generalization performance. Unlike instance based sampling, task based sampling is much more challenging due to the implicit definition of the task in each episode. Therefore, we accordingly propose a greedy class-pair based sampling method, which selects difficult tasks according to class-pair potentials. We evaluate our adaptive task sampling method on two few-shot classification benchmarks, and it achieves consistent improvements across different feature backbones, meta-learning algorithms and datasets., Comment: ECCV2020

Image Super-Resolution (SR) is an important class of image processing techniques to enhance the resolution of images and videos in computer vision. Recent years have witnessed remarkable progress of image super-resolution using deep learning techniques. This article aims to provide a comprehensive survey on recent advances of image super-resolution using deep learning approaches. In general, we can roughly group the existing studies of SR techniques into three major categories: supervised SR, unsupervised SR, and domain-specific SR. In addition, we also cover some other important issues, such as publicly available benchmark datasets and performance evaluation metrics. Finally, we conclude this survey by highlighting several future directions and open issues which should be further addressed by the community in the future., Comment: Accepted by IEEE TPAMI

Origami designs offer extreme reconfigurability due to hinge rotation and facet deformation. This can be exploited to make lightweight metamaterials with controlled deployability and tunable properties. Here, we create a family of origami-inspired cellular metamaterials which can be programmed to have various stability characteristics and mechanical responses in three independent orthogonal directions. The cellular metamaterials were constructed from their origami unit cell that can have one or two admissible closed-loop configurations. The presence of the second closed-loop configuration leads to the emergence of bi-stability in the cellular metamaterial. We show that the stability and reconfigurability of the origami unit cell, and thus the constructed cellular metamaterials, can be programmed by manipulating the characteristic angles inherited from the origami pattern. Two examples of such programmable metamaterial with bi-stability in out-of-plane direction and anisotropic multi-stability in orthogonal directions are presented. Our study provides a platform to design programmable three-dimensional metamaterials significantly broadening the application envelope of origami.