GSSF: Generalized Structural Sparse Function for Deep Cross-Modal Metric Learning.

Learnable dynamic margin in deep metric learning

Person re-identification remains a critical conundrum of intelligent video surveillance system, which intends to query pedestrians with the same identity. It is of great significance to extract features with sharp discrimination and robustness, especially for images captured in complex scenes, such as illumination variance, different pose and viewpoints. In this paper, we propose a normalized distance aggregation strategy, which integrates three discriminative pedestrian feature extraction models, i.e., low-level local maximal occurrence (LOMO), mid-level salient color names based color descriptor (SCNCD), and high-level feature fusion net (FFN). We utilize each feature descriptor to train two distance metric learning models, including large-scale similarity learning (LSSL) and cross-view quadratic discriminant analysis (XQDA), respectively. Accordingly, the ultimate distance is calculated as the sum of six optimized distance metrics by the min-max normalization with respective weights. In experiments, comprehensive evaluation results indicate that our framework gains superior performance with robustness to illumination, viewpoint as well as pose changes, and exceeds the existing methods on the public dataset VIPeR.

Person re-identification (re-id) usually refers to matching people across disjoint camera views. Many existing methods focus on extracting discriminative features or learning distance metrics to make the intraclass distance smaller than interclass distances. These methods subconsciously assume that pedestrian images are well aligned. However, one major challenge in person re-id is the unconstrained spatial misalignment between image pairs due to view angle changes and pedestrian pose variations. To address this problem, in this paper, we propose Recurrent Matching Network of Spatial Alignment Learning (RMN-SAL) to simulate the human vision perception. Reinforcement learning is introduced to locate attention regions, since it provides a flexible learning strategy for sequential decision-making. A linear mapping is employed to convert the environment state into spatial constraint, comprising spatial alignment into feature learning. And recurrent models are used to extract information from a sequence of corresponding regions. Finally, person re-id is performed based on the global features and the features from the learned alignment regions. Our contributions are: 1) the recurrent matching network, which can subtly combine local feature learning and sequential spatial correspondence learning into an end-to-end framework; 2) the design of a location network, which is based on reinforcement learning and aims to learn task-specific sequential spatial correspondences for different image pairs through the local pairwise internal representation interactions. The proposed model is evaluated on three benchmarks, including Market-1501, DukeMTMC-reID and CUHK03, and achieves better performances than other methods.

Membership inference (MI) attacks pose significant privacy threats to machine learning models by determining whether specific data points were used during training. Existing MI methods rely on static heuristics and simple distance metrics that fail to capture the complex, nonlinear output manifolds of modern deep networks, where member and non-member distributions often overlap. We introduce DynaMI, a novel framework that addresses these limitations through two key innovations: (1) adaptive manifold projection using UMAP to better preserve discriminative information in latent space, and (2) counterfactual reasoning that actively probes membership status by analyzing how controlled perturbations affect sample classification. Unlike traditional approaches, DynaMI dynamically adapts to local output space geometry and operates effectively in blind inference scenarios without requiring shadow models or handcrafted thresholds. Our framework demonstrates superior performance across black-box and gray-box threat models, offering a more robust and semantically aware approach to membership inference in complex neural networks. DynaMI is evaluated across 8 diverse datasets using precision, recall, and F1-score metrics under both label-free and label-aware attack scenarios. Testing includes four non-member generation strategies and multiple latent-space kernels to assess framework robustness and adaptability. Performance is benchmarked against state-of-the-art defenses including MemGuard, MMD+Mixup, Adversarial Regularization, and DP-Adam.

Shows that systems built on a simple statistical technique and a large training database can be automatically optimized to produce classification accuracies of 99% in the domain of handwritten digits. It is also shown that the performance of these systems scale consistently with the size of the training database, where the error rate is cut by more than half for every tenfold increase in the size of the training set from 10 to 100,000 examples. Three distance metrics for the standard nearest neighbor classification system are investigated: a simple Hamming distance metric, a pixel distance metric, and a metric based on the extraction of penstroke features. Systems employing these metrics were trained and tested on a standard, publicly available, database of nearly 225,000 digits provided by the National Institute of Standards and Technology. Additionally, a confidence metric is both introduced by the authors and also discovered and optimized by the system. The new confidence measure proves to be superior to the commonly used nearest neighbor distance. >

Person re-identification is one of the hot topics in computer vision. How to design a robust feature representation to identify pedestrians is a key problem for person re-identification. In this paper, a feature representation based on Multi-Statistics Cascade on Pyramid (MSCP) is proposed for person re-identification. The MSCP feature is composed of deep PCA network feature and hand-crafted features of Local Maximal Occurrence (LOMO) feature and color correlogram. MSCP can characterize the pedestrian images precisely from both global and local views. The Cross-view Quadratic Discriminant Analysis (XQDA) is employed to learn the distance metric of MSCP features. And then a novel re-identification method based on MSCP and XQDA is achieved. Experimental results on VIPeR Dataset demonstrate that our proposed method can achieve superior identification performance compared with six state-of-art methods.

Person re-identification (Re-ID) is a critical task in video surveillance and security systems, aiming to match individuals across non-overlapping camera views. However, significant challenges arise due to variations in illumination, pose, and viewpoint. In this work, we propose a distance fusion-based approach to improve person Re-ID accuracy by effectively combining multiple feature-based distance metrics. We extract discriminative features using Local Maximal Occurrence (LOMO), Gaussian of Gaussian (GOG), and Convolutional Neural Networks (CNN). The extracted features are projected into a lower-dimensional space using Cross-view Quadratic Discriminant Analysis (XQDA), and their distances are computed and normalized using Min-Max scaling. To enhance performance, we implement two fusion strategies: Distance Fusion by Simple Sum (DFSS), an unsupervised method that aggregates normalized distances, and Distance Fusion by Logistic Regression (DFLR), a supervised approach that learns an optimal combination of distances. The proposed techniques are evaluated on four benchmark datasets: VIPeR, PRID450S, GRID, and CUHK01. Experimental results demonstrate that our fusion-based methods significantly outperform individual feature-based approaches, achieving state-of-the-art accuracy. Specifically, the fusion of CNN, LOMO, and GOG in the DFLR framework achieves the best performance across all datasets, with improvements in Rank-1 and Rank-20 accuracy. These findings highlight the effectiveness of distance fusion in improving person Re-ID performance, making it a promising approach for real-world applications.

Person Re-Identification is all about determining a person's entire course as s/he walks around camera-equipped zones. More precisely, person Re-ID is the problem of matching human identities captured from non-overlapping surveillance cameras. In this work, we propose an approach that learns a new low-dimensional metric space in an attempt to cut down multi-camera matching errors. We represent the training and test samples by concatenating handcrafted features. Then, the method performs a two-step ranking using elementary distance metrics and followed by an ensemble of weighted binary classifiers. We validate our approach on CUHK01 and PRID450s datasets, providing only a sample per class for probe and only a sample for gallery (single-shot). According to the experiments, our method achieves CMC Rank-1 results up to 61.1 and 75.4, following leading literature protocols, for CUHK01 and PRID450s, respectively.

Unsupervised methods for learning distributed representations of words are ubiquitous in today's NLP research, but far less is known about the best ways to learn distributed phrase or sentence representations from unlabelled data. This paper is a systematic comparison of models that learn such representations. We find that the optimal approach depends critically on the intended application. Deeper, more complex models are preferable for representations to be used in supervised systems, but shallow log-linear models work best for building representation spaces that can be decoded with simple spatial distance metrics. We also propose two new unsupervised representation-learning objectives designed to optimise the trade-off between training time, domain portability and performance.

In plenty of scenarios, data can be represented as vectors and then mathematically abstracted as points in a Euclidean space. Because a great number of machine learning and data mining applications need proximity measures over data, a simple and universal distance metric is desirable, and metric learning methods have been explored to produce sensible distance measures consistent with data relationship. However, most existing methods suffer from limited labeled data and expensive training. In this paper, we address these two issues through employing abundant unlabeled data and pursuing sparsity of metrics, resulting in a novel metric learning approach called semi-supervised sparse metric learning. Two important contributions of our approach are: 1) it propagates scarce prior affinities between data to the global scope and incorporates the full affinities into the metric learning; and 2) it uses an efficient alternating linearization method to directly optimize the sparse metric. Compared with conventional methods, ours can effectively take advantage of semi-supervision and automatically discover the sparse metric structure underlying input data patterns. We demonstrate the efficacy of the proposed approach with extensive experiments carried out on six datasets, obtaining clear performance gains over the state-of-the-arts.

Recent studies show that neural natural processing models for medical code prediction suffer from a label imbalance issue. This study aims to investigate further imbalance in a medical code prediction dataset in terms of demographic variables and analyse performance differences in demographic groups. We use sample-based metrics to correctly evaluate the performance in terms of the data subject. Also, a simple label distance metric is proposed to quantify the difference in the label distribution between a group and the entire data. Our analysis results reveal that the model performs differently towards different demographic groups: significant differences between age groups and between insurance types are observed. Interestingly, we found a weak positive correlation between the number of training data of the group and the performance of the group. However, a strong negative correlation between the label distance of the group and the performance of the group is observed. This result suggests that the model tends to perform poorly in the group whose label distribution is different from the global label distribution of the training data set. Further analysis of the model performance is required to identify the cause of these differences and to improve the model building.

Person Re-Identification aims to retrieve person identities from images captured by multiple cameras or the same cameras in different time instances and locations. Because of its importance in many vision applications from surveillance to human-machine interaction, person re-identification methods need to be reliable and fast. While more and more deep architectures are proposed for increasing performance, those methods also increase overall model complexity. This paper proposes a lightweight network that combines global, part-based, and channel features in a unified multi-branch architecture that builds on the resource-efficient OSNet backbone. Using a well-founded combination of training techniques and design choices, our final model achieves state-of-the-art results on CUHK03 labeled, CUHK03 detected, and Market-1501 with 85.1% mAP / 87.2% rank1, 82.4% mAP / 84.9% rank1, and 91.5% mAP / 96.3% rank1, respectively.

We consider the problem of deriving compressed perceptual representation of multivariate time series and using it for efficient indexing and similarity search. Our algorithm is based on the identification of perceptual skeletons in multidimensional space and the use of these "simplifications" in similarity measurements. We illustrate the performance of the algorithm in a financial modeling application. Our results indicate that the skeleton representation outperforms the traditional approaches and is robust enough to be used even with the simplest distance metrics.

This paper proposes a novel approach to person re-identification, a fundamental task in distributed multi-camera surveillance systems. Although a variety of powerful algorithms have been presented in the past few years, most of them usually focus on designing hand-crafted features and learning metrics either individually or sequentially. Different from previous works, we formulate a unified deep ranking framework that jointly tackles both of these key components to maximize their strengths. We start from the principle that the correct match of the probe image should be positioned in the top rank within the whole gallery set. An effective learning-to-rank algorithm is proposed to minimize the cost corresponding to the ranking disorders of the gallery. The ranking model is solved with a deep convolutional neural network (CNN) that builds the relation between input image pairs and their similarity scores through joint representation learning directly from raw image pixels. The proposed framework allows us to get rid of feature engineering and does not rely on any assumption. An extensive comparative evaluation is given, demonstrating that our approach significantly outperforms all the state-of-the-art approaches, including both traditional and CNN-based methods on the challenging VIPeR, CUHK-01, and CAVIAR4REID datasets. In addition, our approach has better ability to generalize across datasets without fine-tuning.

Despite being often considered less challenging than image-based person re-identification (re-id), video-based person re-id is still appealing as it mimics a more realistic scenario owing to the availability of pedestrian sequences from surveillance cameras. In order to exploit the temporal information provided, a number of feature extraction methods have been proposed. Although the features could be equally learned at a significantly higher computational cost, the scarce nature of labelled re-id datasets encourages the development of robust hand-crafted feature representations as an efficient alternative, especially when novel distance metrics or multi-shot ranking algorithms are to be validated. This paper presents a novel hand-crafted feature representation for video-based person re-id based on a 3-dimensional hierarchical Gaussian descriptor. Compared to similar approaches, the proposed descriptor (i) does not require any walking cycle extraction, hence avoiding the complexity of this task, (ii) can be easily fed into off-shelf learned distance metrics, (iii) and consistently achieves superior performance regardless of the matching method adopted. The performance of the proposed method was validated on PRID2011 and iLIDS-VID datasets outperforming similar methods on both benchmarks.