The goal of person reidentification (Re-ID) is to identify a given pedestrian from a network of nonoverlapping surveillance cameras. Most existing works follow the supervised learning paradigm which requires pairwise labeled training data for each pair of cameras. However, this limits their scalability to real-world applications where abundant unlabeled data are available. To address this issue, we propose a multi-feature fusion with adaptive graph learning model for unsupervised Re-ID. Our model aims to negotiate comprehensive assessment on the consistent graph structure of pedestrians with the help of special information of feature descriptors. Specifically, we incorporate multi-feature dictionary learning and adaptive multi-feature graph learning into a unified learning model such that the learned dictionaries are discriminative and the subsequent graph structure learning is accurate. An alternating optimization algorithm with proved convergence is developed to solve the final optimization objective. Extensive experiments on four benchmark data sets demonstrate the superiority and effectiveness of the proposed method.

Recognizing human actions in video clips has been an important topic in computer vision. Sufficient labeled data is one of the prerequisites for the good performance of action recognition algorithms. However, while abundant videos can be collected from the Internet, categorizing each video clip is tedious and even time-consuming. Active learning is one way to alleviate the labeling labor by allowing the classifier to choose the most informative unlabeled instances for manual annotation. Among various active learning algorithms, uncertainty sampling is arguably the most widely-used strategy. Conventional uncertainty sampling strategies such as entropy-based methods are usually tested under accuracy. However, in action recognition Average Precision (AP) is an acknowledged evaluation metric, which is somehow ignored in the active learning community. It is defined as the area under the precision-recall curve. In this paper, we propose a novel uncertainty sampling algorithm for action recognition using expected AP. We conduct experiments on three real-world action recognition datasets and show that our algorithm outperforms other uncertainty-based active learning algorithms.

Cold-start recommendation is one of the most challenging problems in recommender systems. An important approach to cold-start recommendation is to conduct an interview for new users, called the interview-based approach . Among the interview-based methods, Representative-Based Matrix Factorization (RBMF) [24] provides an effective solution with appealing merits: it represents users over selected representative items, which makes the recommendations highly intuitive and interpretable. However, RBMF only utilizes a global set of representative items to model all users. Such a representation is somehow too strict and may not be flexible enough to capture varying users’ interests. To address this problem, we propose a novel interview-based model to dynamically create meaningful user groups using decision trees and then select local representative items for different groups. A two-round interview is performed for a new user. In the first round, l 1 global questions are issued for group division, while in the second round, l 2 local-group-specific questions are given to derive local representation. We collect the feedback on the (l 1 +l 2 ) items to learn the user representations. By putting these steps together, we develop a joint optimization model, named local representative-based matrix factorization , for new user recommendations. Extensive experiments on three public datasets have demonstrated the effectiveness of the proposed model compared with several competitive baselines.

Active learning is a machine learning technique that trains a classifier after selecting a subset from an unlabeled dataset for labeling and using the selected data for training. Recently, batch mode active learning, which selects a batch of samples to label in parallel, has attracted a lot of attention. Its challenge lies in the choice of criteria used for guiding the search of the optimal batch. In this paper, we propose a novel approach to selecting the optimal batch of queries by minimizing the α-relative Pearson divergence (RPE) between the labeled and the original datasets. This particular divergence is chosen since it can distinguish the optimal batch more easily than other measures especially when available candidates are similar. The proposed objective is a min-max optimization problem, and it is difficult to solve due to the involvement of both minimization and maximization. We find that the objective has an equivalent convex form, and thus a global optimal solution can be obtained. Then the subgradient method can be applied to solve the simplified convex problem. Our empirical studies on UCI datasets demonstrate the effectiveness of the proposed approach compared with the state-of-the-art batch mode active learning methods.

In classification tasks, labeled data is a necessity but sometimes difficult or expensive to obtain. On the contrary, unlabeled data is usually abundant. Recently, different active learning algorithms are proposed to alleviate this issue by selecting the most informative data points to label. One family of active learning methods comes from Optimum Experimental Design (OED) in statistics. Instead of selecting data points one by one iteratively, OED-based approaches select data in a one-shot manner, that is, a fixed-sized subset is selected from the unlabeled dataset for manually labeling. These methods usually use kernels to represent pair-wise similarities between different data points. It is well known that choosing optimal kernel types (e.g. Gaussian kernel) and kernel parameters (e.g. kernel width) is tricky, and a common way to resolve it is by Multiple Kernel Learning (MKL), i.e., to construct a few candidate kernels and merge them to form a consensus kernel. There would be different ways to combine multiple kernels, one of which, called the the globalised approach is to assign a weight to each candidate kernel. In practice different data points in the same candidate kernel may not have the same contribution in the consensus kernel, this requires assigning different weights to different data points in the same candidate kernel, leading to the localized approach. In this paper, we introduce MKL to OED-based active learning, specifically we propose globalised and localized multiple kernel active learning methods, respectively. Our experiments on six benchmark datasets demonstrate that the proposed methods have better performance than existing OED-based active learning methods.

We propose an unsupervised approach to constructing templates from a large collection of semantic category names, and use the templates as the semantic representation of categories.The main challenge is that many terms have multiple meanings, resulting in a lot of wrong templates.Statistical data and semantic knowledge are extracted from a web corpus to improve template generation.A nonlinear scoring function is proposed and demonstrated to be effective.Experiments show that our approach achieves significantly better results than baseline methods.As an immediate application, we apply the extracted templates to the cleaning of a category collection and see promising results (precision improved from 81% to 89%).

In this paper, we consider the problem of unsupervised feature selection. Recently, spectral feature selection algorithms, which leverage both graph Laplacian and spectral regression, have received increasing attention. However, existing spectral feature selection algorithms suffer from two major problems: 1) since the graph Laplacian is constructed from the original feature space, noisy and irrelevant features may have adverse effect on the estimated graph Laplacian and hence degenerate the quality of the induced graph embedding, 2) since the cluster labels are discrete in natural, relaxing and approximating these labels into a continuous embedding can inevitably introduce noise into the estimated cluster labels. Without considering the noise in the cluster labels, the feature selection process may be misguided. In this paper, we propose a Robust Spectral learning framework for unsupervised Feature Selection (RSFS), which jointly improves the robustness of graph embedding and sparse spectral regression. Compared with existing methods which are sensitive to noisy features, our proposed method utilizes a robust local learning method to construct the graph Laplacian and a robust spectral regression method to handle the noise on the learned cluster labels. In order to solve the proposed optimization problem, an efficient iterative algorithm is proposed. We also show the close connection between the proposed robust spectral regression and robust Huber M-estimator. Experimental results on different datasets show the superiority of RSFS.