Wenqi Wang

ByteDance Seed, Jiaze Chen, Tiantian Fan et al. · bytedance

We introduce Seed1.5-Thinking, capable of reasoning through thinking before responding, resulting in improved performance on a wide range of benchmarks. Seed1.5-Thinking achieves 86.7 on AIME 2024, 55.0 on Codeforces and 77.3 on GPQA, demonstrating excellent reasoning abilities in STEM and coding. Beyond reasoning tasks, the method demonstrates notable generalization across diverse domains. For instance, it surpasses DeepSeek R1 by 8% in win rate on non-reasoning tasks, indicating its broader applicability. Compared to other state-of-the-art reasoning models, Seed1.5-Thinking is a Mixture-of-Experts (MoE) model with a relatively small size, featuring 20B activated and 200B total parameters. As part of our effort to assess generalized reasoning, we develop two internal benchmarks, BeyondAIME and Codeforces, both of which will be publicly released to support future research. Model trial link: https://www.volcengine.com/experience/ark.

Xiaohong Liu, Xiongkuo Min, Wei Sun et al. · eth-zurich

This paper reports on the NTIRE 2023 Quality Assessment of Video Enhancement Challenge, which will be held in conjunction with the New Trends in Image Restoration and Enhancement Workshop (NTIRE) at CVPR 2023. This challenge is to address a major challenge in the field of video processing, namely, video quality assessment (VQA) for enhanced videos. The challenge uses the VQA Dataset for Perceptual Video Enhancement (VDPVE), which has a total of 1211 enhanced videos, including 600 videos with color, brightness, and contrast enhancements, 310 videos with deblurring, and 301 deshaked videos. The challenge has a total of 167 registered participants. 61 participating teams submitted their prediction results during the development phase, with a total of 3168 submissions. A total of 176 submissions were submitted by 37 participating teams during the final testing phase. Finally, 19 participating teams submitted their models and fact sheets, and detailed the methods they used. Some methods have achieved better results than baseline methods, and the winning methods have demonstrated superior prediction performance.

Shengao Wang, Wenqi Wang, Zecheng Wang et al.

Early children's developmental trajectories set up a natural goal for sample-efficient pretraining of vision foundation models. We introduce BabyVLM-V2, a developmentally grounded framework for infant-inspired vision-language modeling that extensively improves upon BabyVLM-V1 through a longitudinal, multifaceted pretraining set, a versatile model, and, most importantly, DevCV Toolbox for cognitive evaluation. The pretraining set maximizes coverage while minimizing curation of a longitudinal, infant-centric audiovisual corpus, yielding video-utterance, image-utterance, and multi-turn conversational data that mirror infant experiences. DevCV Toolbox adapts all vision-related measures of the recently released NIH Baby Toolbox into a benchmark suite of ten multimodal tasks, covering spatial reasoning, memory, and vocabulary understanding aligned with early children's capabilities. Experimental results show that a compact model pretrained from scratch can achieve competitive performance on DevCV Toolbox, outperforming GPT-4o on some tasks. We hope the principled, unified BabyVLM-V2 framework will accelerate research in developmentally plausible pretraining of vision foundation models.

Ze Lu, Yalei Lv, Wenqi Wang et al.

Deep generative approaches have obtained great success in image inpainting recently. However, most generative inpainting networks suffer from either over-smooth results or aliasing artifacts. The former lacks high-frequency details, while the latter lacks semantic structure. To address this issue, we propose an effective Frequency-Spatial Complementary Network (FSCN) by exploiting rich semantic information in both spatial and frequency domains. Specifically, we introduce an extra Frequency Branch and Frequency Loss on the spatial-based network to impose direct supervision on the frequency information, and propose a Frequency-Spatial Cross-Attention Block (FSCAB) to fuse multi-domain features and combine the corresponding characteristics. With our FSCAB, the inpainting network is capable of capturing frequency information and preserving visual consistency simultaneously. Extensive quantitative and qualitative experiments demonstrate that our inpainting network can effectively achieve superior results, outperforming previous state-of-the-art approaches with significantly fewer parameters and less computation cost. The code will be released soon.

Chenwei Lou, Zewei Sun, Xinnian Liang et al.

Large Language Models (LLMs) have demonstrated remarkable capabilities but often face challenges with tasks requiring sophisticated reasoning. While Chain-of-Thought (CoT) prompting significantly enhances reasoning, it indiscriminately generates lengthy reasoning steps for all queries, leading to substantial computational costs and inefficiency, especially for simpler inputs. To address this critical issue, we introduce AdaCoT (Adaptive Chain-of-Thought), a novel framework enabling LLMs to adaptively decide when to invoke CoT. AdaCoT framed adaptive reasoning as a Pareto optimization problem that seeks to balance model performance with the costs associated with CoT invocation (both frequency and computational overhead). We propose a reinforcement learning (RL) based method, specifically utilizing Proximal Policy Optimization (PPO), to dynamically control the CoT triggering decision boundary by adjusting penalty coefficients, thereby allowing the model to determine CoT necessity based on implicit query complexity. A key technical contribution is Selective Loss Masking (SLM), designed to counteract decision boundary collapse during multi-stage RL training, ensuring robust and stable adaptive triggering. Experimental results demonstrate that AdaCoT successfully navigates the Pareto frontier, achieving substantial reductions in CoT usage for queries not requiring elaborate reasoning. For instance, on our production traffic testset, AdaCoT reduced CoT triggering rates to as low as 3.18\% and decreased average response tokens by 69.06%, while maintaining high performance on complex tasks.

Wenqi Wang, Reuben Tan, Pengyue Zhu et al.

Spatial intelligence (SI) represents a cognitive ability encompassing the visualization, manipulation, and reasoning about spatial relationships, underpinning disciplines from neuroscience to robotics. We introduce SITE, a benchmark dataset towards SI Thorough Evaluation in a standardized format of multi-choice visual question-answering, designed to assess large vision-language models' spatial intelligence across diverse visual modalities (single-image, multi-image, and video) and SI factors (figural to environmental scales, spatial visualization and orientation, intrinsic and extrinsic, static and dynamic). Our approach to curating the benchmark combines a bottom-up survey about 31 existing datasets and a top-down strategy drawing upon three classification systems in cognitive science, which prompt us to design two novel types of tasks about view-taking and dynamic scenes. Extensive experiments reveal that leading models fall behind human experts especially in spatial orientation, a fundamental SI factor. Moreover, we demonstrate a positive correlation between a model's spatial reasoning proficiency and its performance on an embodied AI task.

Ling He, Yanxin Chen, Wenqi Wang et al.

This study employs cutting-edge wearable monitoring technology to conduct high-precision, high-temporal-resolution (1-second interval) cognitive load assessment on electroencephalogram (EEG) data from the FP1 channel and heart rate variability (HRV) data of secondary vocational students. By jointly analyzing these two critical physiological indicators, the research delves into their application value in assessing cognitive load among secondary vocational students and their utility across various tasks. The study designed two experiments to validate the efficacy of the proposed approach: Initially, a random forest classification model, developed using the N-BACK task, enabled the precise decoding of physiological signal characteristics in secondary vocational students under different levels of cognitive load, achieving a classification accuracy of 97%. Subsequently, this classification model was applied in a cross-task experiment involving the National Computer Rank Examination (Level-1), demonstrating the method's significant applicability and cross-task transferability in diverse learning contexts. Conducted with high portability, this research holds substantial theoretical and practical significance for optimizing teaching resource allocation in secondary vocational education, as well as for cognitive load assessment methods and monitoring. Currently, the research findings are undergoing trial implementation in the school.

Wenqi Wang, Jacob Bieker, Rossella Arcucci et al.

In recent years, the convergence of data-driven machine learning models with Data Assimilation (DA) offers a promising avenue for enhancing weather forecasting. This study delves into this emerging trend, presenting our methodologies and outcomes. We harnessed the UK's local ERA5 850 hPa temperature data and refined the U-STN12 global weather forecasting model, tailoring its predictions to the UK's climate nuances. From the ASOS network, we sourced T2m data, representing ground observations across the UK. We employed the advanced kriging method with a polynomial drift term for consistent spatial resolution. Furthermore, Gaussian noise was superimposed on the ERA5 T850 data, setting the stage for ensuing multi-time step synthetic observations. Probing into the assimilation impacts, the ASOS T2m data was integrated with the ERA5 T850 dataset. Our insights reveal that while global forecast models can adapt to specific regions, incorporating atmospheric data in DA significantly bolsters model accuracy. Conversely, the direct assimilation of surface temperature data tends to mitigate this enhancement, tempering the model's predictive prowess.

Wenlin Wang, Hongteng Xu, Zhe Gan et al.

We propose a novel graph-driven generative model, that unifies multiple heterogeneous learning tasks into the same framework. The proposed model is based on the fact that heterogeneous learning tasks, which correspond to different generative processes, often rely on data with a shared graph structure. Accordingly, our model combines a graph convolutional network (GCN) with multiple variational autoencoders, thus embedding the nodes of the graph i.e., samples for the tasks) in a uniform manner while specializing their organization and usage to different tasks. With a focus on healthcare applications (tasks), including clinical topic modeling, procedure recommendation and admission-type prediction, we demonstrate that our method successfully leverages information across different tasks, boosting performance in all tasks and outperforming existing state-of-the-art approaches.

Wenlin Wang, Hongteng Xu, Ruiyi Zhang et al.

Traditional collaborative filtering (CF) based recommender systems tend to perform poorly when the user-item interactions/ratings are highly scarce. To address this, we propose a learning framework that improves collaborative filtering with a synthetic feedback loop (CF-SFL) to simulate the user feedback. The proposed framework consists of a "recommender" and a "virtual user". The "recommender" is formulated as a CF model, recommending items according to observed user preference. The "virtual user" estimates rewards from the recommended items and generates a \emph{feedback} in addition to the observed user preference. The "recommender" connected with the "virtual user" constructs a closed loop, that recommends users with items and imitates the \emph{unobserved} feedback of the users to the recommended items. The synthetic feedback is used to augment the observed user preference and improve recommendation results. Theoretically, such model design can be interpreted as inverse reinforcement learning, which can be learned effectively via rollout (simulation). Experimental results show that the proposed framework is able to enrich the learning of user preference and boost the performance of existing collaborative filtering methods on multiple datasets.

Wenlin Wang, Hongteng Xu, Guoyin Wang et al.

We propose an optimal transport (OT) framework for generalized zero-shot learning (GZSL), seeking to distinguish samples for both seen and unseen classes, with the assist of auxiliary attributes. The discrepancy between features and attributes is minimized by solving an optimal transport problem. {Specifically, we build a conditional generative model to generate features from seen-class attributes, and establish an optimal transport between the distribution of the generated features and that of the real features.} The generative model and the optimal transport are optimized iteratively with an attribute-based regularizer, that further enhances the discriminative power of the generated features. A classifier is learned based on the features generated for both the seen and unseen classes. In addition to generalized zero-shot learning, our framework is also applicable to standard and transductive ZSL problems. Experiments show that our optimal transport-based method outperforms state-of-the-art methods on several benchmark datasets.

Kai Wang, Fuyuan Shi, Wenqi Wang et al.

This paper presents an improved scheme for the generation and adaption of synthetic images for the training of deep Convolutional Neural Networks(CNNs) to perform the object detection task in smart vending machines. While generating synthetic data has proved to be effective for complementing the training data in supervised learning methods, challenges still exist for generating virtual images which are similar to those of the complex real scenes and minimizing redundant training data. To solve these problems, we consider the simulation of cluttered objects placed in a virtual scene and the wide-angle camera with distortions used to capture the whole scene in the data generation process, and post-processed the generated images with a elaborately-designed generative network to make them more similar to the real images. Various experiments have been conducted to prove the efficiency of using the generated virtual images to enhance the detection precision on existing datasets with limited real training data and the generalization ability of applying the trained network to datasets collected in new environment.

Wenqi Wang, Run Wang, Lina Wang et al.

Deep neural networks (DNNs) have achieved remarkable success in various tasks (e.g., image classification, speech recognition, and natural language processing (NLP)). However, researchers have demonstrated that DNN-based models are vulnerable to adversarial examples, which cause erroneous predictions by adding imperceptible perturbations into legitimate inputs. Recently, studies have revealed adversarial examples in the text domain, which could effectively evade various DNN-based text analyzers and further bring the threats of the proliferation of disinformation. In this paper, we give a comprehensive survey on the existing studies of adversarial techniques for generating adversarial texts written by both English and Chinese characters and the corresponding defense methods. More importantly, we hope that our work could inspire future studies to develop more robust DNN-based text analyzers against known and unknown adversarial techniques. We classify the existing adversarial techniques for crafting adversarial texts based on the perturbation units, helping to better understand the generation of adversarial texts and build robust models for defense. In presenting the taxonomy of adversarial attacks and defenses in the text domain, we introduce the adversarial techniques from the perspective of different NLP tasks. Finally, we discuss the existing challenges of adversarial attacks and defenses in texts and present the future research directions in this emerging and challenging field.

Wenqi Wang, Vaneet Aggarwal, Shuchin Aeron

Tensor train is a hierarchical tensor network structure that helps alleviate the curse of dimensionality by parameterizing large-scale multidimensional data via a set of network of low-rank tensors. Associated with such a construction is a notion of Tensor Train subspace and in this paper we propose a TT-PCA algorithm for estimating this structured subspace from the given data. By maintaining low rank tensor structure, TT-PCA is more robust to noise comparing with PCA or Tucker-PCA. This is borne out numerically by testing the proposed approach on the Extended YaleFace Dataset B.

Wenqi Wang, Yifan Sun, Brian Eriksson et al.

Deep neural networks have demonstrated state-of-the-art performance in a variety of real-world applications. In order to obtain performance gains, these networks have grown larger and deeper, containing millions or even billions of parameters and over a thousand layers. The trade-off is that these large architectures require an enormous amount of memory, storage, and computation, thus limiting their usability. Inspired by the recent tensor ring factorization, we introduce Tensor Ring Networks (TR-Nets), which significantly compress both the fully connected layers and the convolutional layers of deep neural networks. Our results show that our TR-Nets approach {is able to compress LeNet-5 by $11\times$ without losing accuracy}, and can compress the state-of-the-art Wide ResNet by $243\times$ with only 2.3\% degradation in {Cifar10 image classification}. Overall, this compression scheme shows promise in scientific computing and deep learning, especially for emerging resource-constrained devices such as smartphones, wearables, and IoT devices.

Wenlin Wang, Zhe Gan, Wenqi Wang et al.

We propose a Topic Compositional Neural Language Model (TCNLM), a novel method designed to simultaneously capture both the global semantic meaning and the local word ordering structure in a document. The TCNLM learns the global semantic coherence of a document via a neural topic model, and the probability of each learned latent topic is further used to build a Mixture-of-Experts (MoE) language model, where each expert (corresponding to one topic) is a recurrent neural network (RNN) that accounts for learning the local structure of a word sequence. In order to train the MoE model efficiently, a matrix factorization method is applied, by extending each weight matrix of the RNN to be an ensemble of topic-dependent weight matrices. The degree to which each member of the ensemble is used is tied to the document-dependent probability of the corresponding topics. Experimental results on several corpora show that the proposed approach outperforms both a pure RNN-based model and other topic-guided language models. Further, our model yields sensible topics, and also has the capacity to generate meaningful sentences conditioned on given topics.

Wenqi Wang, Vaneet Aggarwal, Shuchin Aeron

In this paper, we propose a Tensor Train Neighborhood Preserving Embedding (TTNPE) to embed multi-dimensional tensor data into low dimensional tensor subspace. Novel approaches to solve the optimization problem in TTNPE are proposed. For this embedding, we evaluate novel trade-off gain among classification, computation, and dimensionality reduction (storage) for supervised learning. It is shown that compared to the state-of-the-arts tensor embedding methods, TTNPE achieves superior trade-off in classification, computation, and dimensionality reduction in MNIST handwritten digits and Weizmann face datasets.

Wenqi Wang, Vaneet Aggarwal, Shuchin Aeron

Using the matrix product state (MPS) representation of the recently proposed tensor ring decompositions, in this paper we propose a tensor completion algorithm, which is an alternating minimization algorithm that alternates over the factors in the MPS representation. This development is motivated in part by the success of matrix completion algorithms that alternate over the (low-rank) factors. In this paper, we propose a spectral initialization for the tensor ring completion algorithm and analyze the computational complexity of the proposed algorithm. We numerically compare it with existing methods that employ a low rank tensor train approximation for data completion and show that our method outperforms the existing ones for a variety of real computer vision settings, and thus demonstrate the improved expressive power of tensor ring as compared to tensor train.

Wenlin Wang, Changyou Chen, Wenqi Wang et al.

We present Earliness-Aware Deep Convolutional Networks (EA-ConvNets), an end-to-end deep learning framework, for early classification of time series data. Unlike most existing methods for early classification of time series data, that are designed to solve this problem under the assumption of the availability of a good set of pre-defined (often hand-crafted) features, our framework can jointly perform feature learning (by learning a deep hierarchy of \emph{shapelets} capturing the salient characteristics in each time series), along with a dynamic truncation model to help our deep feature learning architecture focus on the early parts of each time series. Consequently, our framework is able to make highly reliable early predictions, outperforming various state-of-the-art methods for early time series classification, while also being competitive when compared to the state-of-the-art time series classification algorithms that work with \emph{fully observed} time series data. To the best of our knowledge, the proposed framework is the first to perform data-driven (deep) feature learning in the context of early classification of time series data. We perform a comprehensive set of experiments, on several benchmark data sets, which demonstrate that our method yields significantly better predictions than various state-of-the-art methods designed for early time series classification. In addition to obtaining high accuracies, our experiments also show that the learned deep shapelets based features are also highly interpretable and can help gain better understanding of the underlying characteristics of time series data.

Wenqi Wang, Vaneet Aggarwal, Shuchin Aeron

In this paper, we consider clustering data that is assumed to come from one of finitely many pointed convex polyhedral cones. This model is referred to as the Union of Polyhedral Cones (UOPC) model. Similar to the Union of Subspaces (UOS) model where each data from each subspace is generated from a (unknown) basis, in the UOPC model each data from each cone is assumed to be generated from a finite number of (unknown) \emph{extreme rays}.To cluster data under this model, we consider several algorithms - (a) Sparse Subspace Clustering by Non-negative constraints Lasso (NCL), (b) Least squares approximation (LSA), and (c) K-nearest neighbor (KNN) algorithm to arrive at affinity between data points. Spectral Clustering (SC) is then applied on the resulting affinity matrix to cluster data into different polyhedral cones. We show that on an average KNN outperforms both NCL and LSA and for this algorithm we provide the deterministic conditions for correct clustering. For an affinity measure between the cones it is shown that as long as the cones are not very coherent and as long as the density of data within each cone exceeds a threshold, KNN leads to accurate clustering. Finally, simulation results on real datasets (MNIST and YaleFace datasets) depict that the proposed algorithm works well on real data indicating the utility of the UOPC model and the proposed algorithm.

Wenqi Wang, Vaneet Aggarwal, Shuchin Aeron

Using the matrix product state (MPS) representation of tensor train decompositions, in this paper we propose a tensor completion algorithm which alternates over the matrices (tensors) in the MPS representation. This development is motivated in part by the success of matrix completion algorithms which alternate over the (low-rank) factors. We comment on the computational complexity of the proposed algorithm and numerically compare it with existing methods employing low rank tensor train approximation for data completion as well as several other recently proposed methods. We show that our method is superior to existing ones for a variety of real settings.

Wenqi Wang, Shuchin Aeron, Vaneet Aggarwal

In this paper we present deterministic conditions for success of sparse subspace clustering (SSC) under missing data, when data is assumed to come from a Union of Subspaces (UoS) model. We consider two algorithms, which are variants of SSC with entry-wise zero-filling that differ in terms of the optimization problems used to find affinity matrix for spectral clustering. For both the algorithms, we provide deterministic conditions for any pattern of missing data such that perfect clustering can be achieved. We provide extensive sets of simulation results for clustering as well as completion of data at missing entries, under the UoS model. Our experimental results indicate that in contrast to the full data case, accurate clustering does not imply accurate subspace identification and completion, indicating the natural order of relative hardness of these problems.

Wenqi Wang, Shuchin Aeron, Vaneet Aggarwal

In this paper we present deterministic analysis of sufficient conditions for sparse subspace clustering under missing data, when data is assumed to come from a Union of Subspaces (UoS) model. In this context we consider two cases, namely Case I when all the points are sampled at the same co-ordinates, and Case II when points are sampled at different locations. We show that results for Case I directly follow from several existing results in the literature, while results for Case II are not as straightforward and we provide a set of dual conditions under which, perfect clustering holds true. We provide extensive set of simulation results for clustering as well as completion of data under missing entries, under the UoS model. Our experimental results indicate that in contrast to the full data case, accurate clustering does not imply accurate subspace identification and completion, indicating the natural order of relative hardness of these problems.