Yi Guo

Yi Guo, Fanliu Kong, Xiaoyang Li et al.

Quantization emerges as one of the most promising compression technologies for deploying efficient large models for various real time application in recent years. Considering that the storage and IO of weights take up the vast majority of the overhead inside a large model, weight only quantization can lead to large gains. However, existing quantization schemes suffer from significant accuracy degradation at very low bits, or require some additional computational overhead when deployed, making it difficult to be applied to large-scale applications in industry. In this paper, we propose decoupleQ, achieving a substantial increase in model accuracy, especially at very low bits. decoupleQ abandons the traditional heuristic quantization paradigm and decouples the model parameters into integer and floating-point parts, thus transforming the quantization problem into a traditional mathematical optimization problem with constraints, which is then solved alternatively by off-the-shelf optimization methods. Quantization via decoupleQ is linear and uniform, making it hardware-friendlier than non-uniform counterpart, and enabling the idea to be migrated to high-bit quantization to enhance its robustness. Our method has achieved well on-line accuracy near fp16/bf16 on the 2-bit quantization of large speech models in ByteDance. The code is available at https://github.com/bytedance/decoupleQ

Yi Guo, Wei Wang, Zhihang Yuan et al.

Generative models like Flow Matching have achieved state-of-the-art performance but are often hindered by a computationally expensive iterative sampling process. To address this, recent work has focused on few-step or one-step generation by learning the average velocity field, which directly maps noise to data. MeanFlow, a leading method in this area, learns this field by enforcing a differential identity that connects the average and instantaneous velocities. In this work, we argue that this differential formulation is a limiting special case of a more fundamental principle. We return to the first principles of average velocity and leverage the additivity property of definite integrals. This leads us to derive a novel, purely algebraic identity we term Interval Splitting Consistency. This identity establishes a self-referential relationship for the average velocity field across different time intervals without resorting to any differential operators. Based on this principle, we introduce SplitMeanFlow, a new training framework that enforces this algebraic consistency directly as a learning objective. We formally prove that the differential identity at the core of MeanFlow is recovered by taking the limit of our algebraic consistency as the interval split becomes infinitesimal. This establishes SplitMeanFlow as a direct and more general foundation for learning average velocity fields. From a practical standpoint, our algebraic approach is significantly more efficient, as it eliminates the need for JVP computations, resulting in simpler implementation, more stable training, and broader hardware compatibility. One-step and two-step SplitMeanFlow models have been successfully deployed in large-scale speech synthesis products (such as Doubao), achieving speedups of 20x.

Namal Jayasuriya, Yi Guo, Wen Hu et al.

Estimation of a single leaf area can be a measure of crop growth and a phenotypic trait to breed new varieties. It has also been used to measure leaf area index and total leaf area. Some studies have used hand-held cameras, image processing 3D reconstruction and unsupervised learning-based methods to estimate the leaf area in plant images. Deep learning works well for object detection and segmentation tasks; however, direct area estimation of objects has not been explored. This work investigates deep learning-based leaf area estimation, for RGBD images taken using a mobile camera setup in real-world scenarios. A dataset for attached leaves captured with a top angle view and a dataset for detached single leaves were collected for model development and testing. First, image processing-based area estimation was tested on manually segmented leaves. Then a Mask R-CNN-based model was investigated, and modified to accept RGBD images and to estimate the leaf area. The detached-leaf data set was then mixed with the attached-leaf plant data set to estimate the single leaf area for plant images, and another network design with two backbones was proposed: one for segmentation and the other for area estimation. Instead of trying all possibilities or random values, an agile approach was used in hyperparameter tuning. The final model was cross-validated with 5-folds and tested with two unseen datasets: detached and attached leaves. The F1 score with 90% IoA for segmentation result on unseen detached-leaf data was 1.0, while R-squared of area estimation was 0.81. For unseen plant data segmentation, the F1 score with 90% IoA was 0.59, while the R-squared score was 0.57. The research suggests using attached leaves with ground truth area to improve the results.

Dai Shi, Andi Han, Yi Guo et al.

Dimensionality reduction (DR) and manifold learning (ManL) have been applied extensively in many machine learning tasks, including signal processing, speech recognition, and neuroinformatics. However, the understanding of whether DR and ManL models can generate valid learning results remains unclear. In this work, we investigate the validity of learning results of some widely used DR and ManL methods through the chart mapping function of a manifold. We identify a fundamental problem of these methods: the mapping functions induced by these methods violate the basic settings of manifolds, and hence they are not learning manifold in the mathematical sense. To address this problem, we provide a provably correct algorithm called fixed points Laplacian mapping (FPLM), that has the geometric guarantee to find a valid manifold representation (up to a homeomorphism). Combining one additional condition(orientation preserving), we discuss a sufficient condition for an algorithm to be bijective for any d-simplex decomposition result on a d-manifold. However, constructing such a mapping function and its computational method satisfying these conditions is still an open problem in mathematics.

Stephen Tierney, Yi Guo, Junbin Gao

In this paper we present Collaborative Low-Rank Subspace Clustering. Given multiple observations of a phenomenon we learn a unified representation matrix. This unified matrix incorporates the features from all the observations, thus increasing the discriminative power compared with learning the representation matrix on each observation separately. Experimental evaluation shows that our method outperforms subspace clustering on separate observations and the state of the art collaborative learning algorithm.

Stephen Tierney, Junbin Gao, Yi Guo et al.

In machine learning it is common to interpret each data point as a vector in Euclidean space. However the data may actually be functional i.e.\ each data point is a function of some variable such as time and the function is discretely sampled. The naive treatment of functional data as traditional multivariate data can lead to poor performance since the algorithms are ignoring the correlation in the curvature of each function. In this paper we propose a tractable method to cluster functional data or curves by adapting the Euclidean Low-Rank Representation (LRR) to the curve manifold. Experimental evaluation on synthetic and real data reveals that this method massively outperforms prior clustering methods in both speed and accuracy when clustering functional data.

Stephen Tierney, Yi Guo, Junbin Gao

Sparse Subspace Clustering (SSC) has been used extensively for subspace identification tasks due to its theoretical guarantees and relative ease of implementation. However SSC has quadratic computation and memory requirements with respect to the number of input data points. This burden has prohibited SSCs use for all but the smallest datasets. To overcome this we propose a new method, k-SSC, that screens out a large number of data points to both reduce SSC to linear memory and computational requirements. We provide theoretical analysis for the bounds of success for k-SSC. Our experiments show that k-SSC exceeds theoretical expectations and outperforms existing SSC approximations by maintaining the classification performance of SSC. Furthermore in the spirit of reproducible research we have publicly released the source code for k-SSC

Ming Yin, Junbin Gao, Shengli Xie et al.

The plenty information from multiple views data as well as the complementary information among different views are usually beneficial to various tasks, e.g., clustering, classification, de-noising. Multi-view subspace clustering is based on the fact that the multi-view data are generated from a latent subspace. To recover the underlying subspace structure, the success of the sparse and/or low-rank subspace clustering has been witnessed recently. Despite some state-of-the-art subspace clustering approaches can numerically handle multi-view data, by simultaneously exploring all possible pairwise correlation within views, the high order statistics is often disregarded which can only be captured by simultaneously utilizing all views. As a consequence, the clustering performance for multi-view data is compromised. To address this issue, in this paper, a novel multi-view clustering method is proposed by using \textit{t-product} in third-order tensor space. Based on the circular convolution operation, multi-view data can be effectively represented by a \textit{t-linear} combination with sparse and low-rank penalty using "self-expressiveness". Our extensive experimental results on facial, object, digits image and text data demonstrate that the proposed method outperforms the state-of-the-art methods in terms of many criteria.

Ming Yin, Shengli Xie, Yi Guo et al.

Due to its promising classification performance, sparse representation based classification(SRC) algorithm has attracted great attention in the past few years. However, the existing SRC type methods apply only to vector data in Euclidean space. As such, there is still no satisfactory approach to conduct classification task for symmetric positive definite (SPD) matrices which is very useful in computer vision. To address this problem, in this paper, a neighborhood preserved kernel SRC method is proposed on SPD manifolds. Specifically, by embedding the SPD matrices into a Reproducing Kernel Hilbert Space (RKHS), the proposed method can perform classification on SPD manifolds through an appropriate Log-Euclidean kernel. Through exploiting the geodesic distance between SPD matrices, our method can effectively characterize the intrinsic local Riemannian geometry within data so as to well unravel the underlying sub-manifold structure. Despite its simplicity, experimental results on several famous database demonstrate that the proposed method achieves better classification results than the state-of-the-art approaches.

Junbin Gao, Yi Guo, Zhiyong Wang

Traditional neural networks assume vectorial inputs as the network is arranged as layers of single line of computing units called neurons. This special structure requires the non-vectorial inputs such as matrices to be converted into vectors. This process can be problematic. Firstly, the spatial information among elements of the data may be lost during vectorisation. Secondly, the solution space becomes very large which demands very special treatments to the network parameters and high computational cost. To address these issues, we propose matrix neural networks (MatNet), which takes matrices directly as inputs. Each neuron senses summarised information through bilinear mapping from lower layer units in exactly the same way as the classic feed forward neural networks. Under this structure, back prorogation and gradient descent combination can be utilised to obtain network parameters efficiently. Furthermore, it can be conveniently extended for multimodal inputs. We apply MatNet to MNIST handwritten digits classification and image super resolution tasks to show its effectiveness. Without too much tweaking MatNet achieves comparable performance as the state-of-the-art methods in both tasks with considerably reduced complexity.

Stephen Tierney, Junbin Gao, Yi Guo et al.

In machine learning it is common to interpret each data point as a vector in Euclidean space. However the data may actually be functional i.e.\ each data point is a function of some variable such as time and the function is discretely sampled. The naive treatment of functional data as traditional multivariate data can lead to poor performance since the algorithms are ignoring the correlation in the curvature of each function. In this paper we propose a method to analyse subspace structure of the functional data by using the state of the art Low-Rank Representation (LRR). Experimental evaluation on synthetic and real data reveals that this method massively outperforms conventional LRR in tasks concerning functional data.

Ming Yin, Yi Guo, Junbin Gao et al.

Sparse subspace clustering (SSC), as one of the most successful subspace clustering methods, has achieved notable clustering accuracy in computer vision tasks. However, SSC applies only to vector data in Euclidean space. As such, there is still no satisfactory approach to solve subspace clustering by ${\it self-expressive}$ principle for symmetric positive definite (SPD) matrices which is very useful in computer vision. In this paper, by embedding the SPD matrices into a Reproducing Kernel Hilbert Space (RKHS), a kernel subspace clustering method is constructed on the SPD manifold through an appropriate Log-Euclidean kernel, termed as kernel sparse subspace clustering on the SPD Riemannian manifold (KSSCR). By exploiting the intrinsic Riemannian geometry within data, KSSCR can effectively characterize the geodesic distance between SPD matrices to uncover the underlying subspace structure. Experimental results on two famous database demonstrate that the proposed method achieves better clustering results than the state-of-the-art approaches.

Stephen Tierney, Yi Guo, Junbin Gao

We propose Ordered Subspace Clustering (OSC) to segment data drawn from a sequentially ordered union of subspaces. Similar to Sparse Subspace Clustering (SSC) we formulate the problem as one of finding a sparse representation but include an additional penalty term to take care of sequential data. We test our method on data drawn from infrared hyper spectral, video and motion capture data. Experiments show that our method, OSC, outperforms the state of the art methods: Spatial Subspace Clustering (SpatSC), Low-Rank Representation (LRR) and SSC.