流形学习与非线性回归结合的头部姿态估计

流形学习的目的是发现非线性数据的内在结构,可用于非线性降维。广义回归网络是人工神经网络的一种,可用于非线性回归。基于流形学习和非线性回归,提出了用于解决头部姿态估计的ManiNLR方法。该方法首先用流形学习对图像数据进行降维,然后用非线性回归的方法将数据映射到线性可分空间,利用非线性回归的结果对人脸的头部姿态进行估计。实验结果表明,ManiNLR算法能够较好地估计图像中的头部姿态,并具有较快的速度和较高的鲁棒性。     

Practical age estimation using deep label distribution learning

Age estimation plays an important role in humancomputer interaction system. The lack of large number of facial images with definite age label makes age estimation algorithms inefficient. Deep label distribution learning (DLDL) which employs convolutional neural networks (CNN) and label distribution learning to learn ambiguity from ground-truth age and adjacent ages, has been proven to outperform current state-of-the-art framework. However, DLDL assumes a rough label distribution which covers all ages for any given age label. In this paper, a more practical label distribution paradigm is proposed: we limit age label distribution that only covers a reasonable number of neighboring ages. In addition, we explore different label distributions to improve the performance of the proposed learning model. We employ CNN and the improved label distribution learning to estimate age. Experimental results show that compared to the DLDL, our method is more effective for facial age recognition.     

稀疏回归和流形学习的无监督特征选择算法

针对无标签高维数据的大量出现,对机器学习中无监督特征选择进行了研究。提出了一种结合自表示相似矩阵和流形学习的无监督特征选择算法。首先,通过数据的自表示性质,构建相似矩阵,结合低维流形能够表示高维数据结构这一流形学习思想,建立一种考虑流形学习的无监督特征选择优化模型。其次,为了保证选择更有用及更稀疏的特征,采用◢l◣▼2,1▽范数对优化模型进行约束,使特征之间相互竞争,消除冗余。进而,通过变量交替迭代对优化模型进行求解,并证明了算法的收敛性。最后,通过与其他几个无监督特征算法在四个数据集上的对比实验,表明所给算法的有效性。     

Kernel ridge regression for out-of-sample mapping in supervised manifold learning

Manifold learning methods for unsupervised nonlinear dimensionality reduction have proven effective in the visualization of high dimensional data sets. When dealing with classification tasks, supervised extensions of manifold learning techniques, in which class labels are used to improve the embedding of the training points, require an appropriate method for out-of-sample mapping.In this paper we propose multi-output kernel ridge regression (KRR) for out-of-sample mapping in supervised manifold learning, in place of general regression neural networks (GRNN) that have been adopted by previous studies on the subject. Specifically, we consider a supervised agglomerative variant of Isomap and compare the performance of classification methods when the out-of-sample embedding is based on KRR and GRNN, respectively. Extensive computational experiments, using support vector machines and k-nearest neighbors as base classifiers, provide statistical evidence that out-of-sample mapping based on KRR consistently dominates its GRNN counterpart, and that supervised agglomerative Isomap with KRR achieves a higher accuracy than direct classification methods on most data sets.     

Transductive Bayesian regression via manifold learning of prior data structure

During the last decades, many studies have been conducted on performing reliable prediction for high-dimensional data that are usually non-linearly correlated with complex patterns. In this paper, we propose a novel Bayesian regression method via non-linear dimensionality reduction. The method incorporates prior information on the underlying structure of original input features to preserve input–output patterns on reduced features, and to provide distributions of predicted values. To verify the effectiveness of the proposed method, we conducted simulations on benchmark and real-world data. Results showed that the method not only better predicts a distribution of forecast estimates compared with other methods, but also more robust and consistent performance on prediction.     

Bone age estimation from carpal radiography images using deep learning

Bone age estimation has been used in medicine to verify whether the bone structure development degree of a person corresponds to their chronological age. Such estimate is useful for prognosis about the development of children and adolescents, as well as for the diagnosis of endocrinological diseases. This work proposes a fully automated methodology for bone age estimation from carpal radiography images. The methodology comprises two steps, the preprocessing of the image and the classification using a convolutional neural network. The system accuracy for different types of preprocessing is evaluated. We compare the accuracy achieved using the full radiography image as input for the neural network and using only parts of the image corresponding to the Phalangeal region, the Epiphyseal region, and the concatenation of these parts with a crop around the wrist. Digital image processing techniques are employed to segment these regions. Experiments are performed using radiography images from the California University Database. The impact of using different pre-trained neural networks for transfer learning is evaluated.     

流形降维最小二乘回归子空间分割

子空间分割方法一直是一种重要的机器学习方法,这些方法在人脸识别和基因表达数据识别等研究中有较好的聚类准确率。然而,这些方法在对高维小样本数据进行聚类时难以取得理想的结果。为了解决这些问题,借鉴流形降维中的局部保持投影法和最小二乘回归子空间分割法,提出流形降维最小二乘回归子空间分割法。该方法通过局部保持投影进行降维,再利用最小二乘回归子空间分割方法实现聚类。在6个生物基因表达数据集和2个图像数据集上的实验表明了该方法的有效性。     

Natural facial expression recognition using differential-AAM and manifold learning

This paper proposes a novel natural facial expression recognition method that recognizes a sequence of dynamic facial expression images using the differential active appearance model (AAM) and manifold learning as follows. First, the differential-AAM features (DAFs) are computed by the difference of the AAM parameters between an input face image and a reference (neutral expression) face image. Second, manifold learning embeds the DAFs on the smooth and continuous feature space. Third, the input facial expression is recognized through two steps: (1) computing the distances between the input image sequence and gallery image sequences using directed Hausdorff distance (DHD) and (2) selecting the expression by a majority voting of k-nearest neighbors (k-NN) sequences in the gallery. The DAFs are robust and efficient for the facial expression analysis due to the elimination of the inter-person, camera, and illumination variations. Since the DAFs treat the neutral expression image as the reference image, the neutral expression image must be found effectively. This is done via the differential facial expression probability density model (DFEPDM) using the kernel density approximation of the positively directional DAFs changing from neutral to angry (happy, surprised) and negatively directional DAFs changing from angry (happy, surprised) to neutral. Then, a face image is considered to be the neutral expression if it has the maximum DFEPDM in the input sequences. Experimental results show that (1) the DAFs improve the facial expression recognition performance over conventional AAM features by 20% and (2) the sequence-based k-NN classifier provides a 95% facial expression recognition performance on the facial expression database (FED06).     

Inductive manifold learning using structured support vector machine

Most manifold learning techniques are used to transform high-dimensional data sets into low-dimensional space. In the use of such techniques, after unseen data samples are added to the data set, retraining is usually necessary. However, retraining is a time-consuming process and no guarantee of the transformation into the exactly same coordinates, thus presenting a barrier to the application of manifold learning as a preprocessing step in predictive modeling. To solve this problem, learning a mapping from high-dimensional representations to low-dimensional coordinates is proposed via structured support vector machine. After training a mapping, low-dimensional representations of unobserved data samples can be easily predicted. Experiments on several datasets show that the proposed method outperforms the existing out-of-sample extension methods.     

Unbalance estimation using linear and nonlinear regression

This paper considers the problem of unbalance estimation of rotating machinery. It is formulated as a parameter estimation problem, where the unknowns enter nonlinearly in a regression model. By use of a certain method, the problem can be reformulated as a linear estimation procedure with a closed form solution. This procedure is sometimes known as the influence coefficient method. In its derivation, no special treatment is devoted to disturbing terms and imperfections in the model. Therefore, a novel method is derived which takes disturbances into account, leading to a nonlinear estimator.The two procedures are compared and analyzed with respect to their statistical accuracy. Using the example of unbalance estimation of a separator, the nonlinear approach is shown to give superior performance.     

Gait analysis for human identification through manifold learning and HMM

With the increasing demands of visual surveillance systems, human identification at a distance has gained more attention from the researchers recently. Gait analysis can be used as an unobtrusive biometric measure to identify people at a distance without any attention of the human subjects. We propose a novel effective method for both automatic viewpoint and person identification by using only the silhouette sequence of the gait. The gait silhouettes are nonlinearly transformed into low-dimensional embedding by Gaussian process latent variable model (GP-LVM), and the temporal dynamics of the gait sequences are modeled by hidden Markov models (HMMs). The experimental results show that our method has higher recognition rate than the other methods.     

Riemannian manifold learning

Recently, manifold learning has been widely exploited in pattern recognition, data analysis, and machine learning. This paper presents a novel framework, called Riemannian manifold learning (RML), based on the assumption that the input high-dimensional data lie on an intrinsically low-dimensional Riemannian manifold. The main idea is to formulate the dimensionality reduction problem as a classical problem in Riemannian geometry, i.e., how to construct coordinate charts for a given Riemannian manifold? We implement the Riemannian normal coordinate chart, which has been the most widely used in Riemannian geometry, for a set of unorganized data points. First, two input parameters (the neighborhood size k and the intrinsic dimension d) are estimated based on an efficient simplicial reconstruction of the underlying manifold. Then, the normal coordinates are computed to map the input high-dimensional data into a low-dimensional space. Experiments on synthetic data as well as real world images demonstrate that our algorithm can learn intrinsic geometric structures of the data, preserve radial geodesic distances, and yield regular embeddings.     

Phoneme recognition using an adaptive supervised manifold learning algorithm

To effectively handle speech data lying on a nonlinear manifold embedded in a high-dimensional acoustic space, in this paper, an adaptive supervised manifold learning algorithm based on locally linear embedding (LLE) for nonlinear dimensionality reduction is proposed to extract the low-dimensional embedded data representations for phoneme recognition. The proposed method aims to make the interclass dissimilarity maximized, while the intraclass dissimilarity minimized in order to promote the discriminating power and generalization ability of the low-dimensional embedded data representations. The performance of the proposed method is compared with five well-known dimensionality reduction methods, i.e., principal component analysis, linear discriminant analysis, isometric mapping (Isomap), LLE as well as the original supervised LLE. Experimental results on three benchmarking speech databases, i.e., the Deterding database, the DARPA TIMIT database, and the ISOLET E-set database, demonstrate that the proposed method obtains promising performance on the phoneme recognition task, outperforming the other used methods.     

Real-time estimation of hand gestures based on manifold learning from monocular videos

Object pose estimation by manifold learning has become a hot research area recently. In this paper, we propose an efficient method that can recover pose and viewpoints for numerous hand gestures from monocular videos based on Locality Preserving Projections. We first select some hand dynamic gestures as primitive hand motions and set a 3D-2D mapping table to relate 3D joint angles of sampling static pose with their projective silhouettes from arbitrary viewpoints. Then the embedding space and explicit mapping function are learnt for every primitive motion. In order to make classification and prediction among those embedding spaces, a Subspace Filtering Algorithm is also proposed which can recognize and recover numerous hand dynamic gestures by the combination of primitive gestures. At last, by using skin color cues and oriented k-Dops, multi-hands can be labeled and tracked separately and accurately. Extensive experimental results demonstrate qualitatively and quantitatively that 3D pose recovery of hands can be achieved by our method robustly and efficiently.     

An elastic manifold learning approach to beat-to-beat interval estimation with ballistocardiography signals

Continuous monitoring of heart rate variation is an important measure to diagnose cardiovascular problems and reduce related morbidity and mortality. The recent advances in wearable sensors have enabled the collection of ballistocardiographic (BCG) records over a long period without sacrificing the user’s normal daily life. However, there are multiple interferences that severely impact the BCG sampling process and thus degrade the signal quality. In this paper, we introduce a novel approach to estimating the beat-to-beat intervals by applying an unsupervised manifold learning framework in a hybrid phase space. First, we map the BCG time series into the three-dimensional space within which the desired BCG sample points are expected to form a low-dimensional manifold. This manifold is then reconstructed by its local linear property to remove the high-frequency noise; and overlapping manifold segments are projected to a low-dimensional principal subspace before aligned to mitigate the low-frequency non-stationary center shifts and amplitude variations. After we take the statistics to analyze the period indicators, the heartbeat intervals can be inferred. The proposed approach was tested with the BCG data collected from 10 subjects in different genders, ages, heights, and weights. We compare the estimates with the ground truth ECG references, and the results show that the proposed algorithm is able to provide reliable and accurate estimates for heart rates and beat-to-beat intervals, with the standard deviation of the interval estimate error of 22 ms.     

Motion estimation using learning automata

Block-matching algorithms (BMAs) are widely employed for motion estimation. BMAs divide input frames into several blocks and minimize an error function for each block to calculate motion vectors. Afterward, each motion vector is applicable for all of the pixels within the block. Since computing the error functions is resource intensive, many fast-search motion estimation algorithms have been suggested to reduce the computational cost. These fast algorithms provide a significant reduction in computation but often converge to a local minimum. A learning automaton is an adaptive decision-making unit that learns the optimal action through repeated interactions with its environment. Learning automata (LA) have been applied successfully to a wide range of applications including pattern recognition, dynamic channel assignment, and social network analysis. In this paper, we apply LA to motion estimation problem, which is one of the basic problems in computer vision. We compare the accuracy and performance of the suggested algorithms with other well-known BMAs. Interestingly, the obtained results indicate high efficiency and accuracy of the proposed methods. The results suggest that simplicity, efficiency, parallel nature, and accuracy of LA-based methods make them a good candidate to solve computer vision problems.     

High-dimensional MRI data analysis using a large-scale manifold learning approach

A novel manifold learning approach is presented to efficiently identify low-dimensional structures embedded in high-dimensional MRI data sets. These low-dimensional structures, known as manifolds, are used in this study for predicting brain tumor progression. The data sets consist of a series of high-dimensional MRI scans for four patients with tumor and progressed regions identified. We attempt to classify tumor, progressed and normal tissues in low-dimensional space. We also attempt to verify if a progression manifold exists—the bridge between tumor and normal manifolds. By identifying and mapping the bridge manifold back to MRI image space, this method has the potential to predict tumor progression. This could be greatly beneficial for patient management. Preliminary results have supported our hypothesis: normal and tumor manifolds are well separated in a low-dimensional space. Also, the progressed manifold is found to lie roughly between the normal and tumor manifolds.     

基于分类与回归混合模型的人脸年龄估计方法

针对现有人脸年龄数据库样本数量少、各年龄段分布不均匀的问题,提出了一种基于分类与回归混合模型的人脸年龄估计方法。该方法主要包含两个方面：特征学习和估计模式。在特征学习方面,利用已有的深度卷积神经网络（CNN）,先在粗糙年龄标注数据集上预训练,再在现有的精确年龄标注数据库上微调,分别得到一个年龄段判别模型和两个年龄估计模型;在估计模式方面,该方法采用由粗到细的策略：首先,将人脸分入青少年、中年、老年和两个重叠区域这五个年龄段;然后,对于青少年和老年采用分类模型估计,对于中年采用回归模型估计,对于重叠区域采用两个模型估计的均值。所提方法在测试集上的平均绝对误差（MAE）为2.56。实验结果表明该方法受不同肤色和性别的影响较小,有较低的误差。     

Categorical structural optimization using discrete manifold learning approach and custom-built evolutionary operators

In present work, we address the non-ordinal categorical design variables, such as different beam/bar cross-section types, various materials or components available within a catalog. We interpret the admissible values of categorical variables as discrete points in multi-dimensional space of physical attributes, which allows computing distances but has no ordering property. Then we propose to use the Isomap manifold learning approach to eliminate the possibly redundant dimensionality and obtain a reduced-order design space in which the geodesic distances are preserved in a low-dimensional graph. Then, taking advantage of the shortest path and the neighbors provided by Dijkstra algorithm, we propose graph-based crossover and mutation operators to be used in evolutionary optimization. The method is applied to the optimal design of truss and frame structures.