Context multi-task visual object tracking via guided filter

Multi-task feature selection with sparse regularization to extract common and task-specific features

In this paper, we formulate particle filter-based object tracking as an exclusive sparse learning problem that exploits contextual information. To achieve this goal, we propose the context-aware exclusive sparse tracker (CEST) to model particle appearances as linear combinations of dictionary templates that are updated dynamically. Learning the representation of each particle is formulated as an exclusive sparse representation problem, where the overall dictionary is composed of multiple group dictionaries that can contain contextual information. With context, CEST is less prone to tracker drift. Interestingly, we show that the popular L1 tracker is a special case of our CEST formulation. The proposed learning problem is efficiently solved using an accelerated proximal gradient method that yields a sequence of closed form updates. To make the tracker much faster, we reduce the number of learning problems to be solved by using the dual problem to quickly and systematically rank and prune particles in each frame. We test our CEST tracker on challenging benchmark sequences that involve heavy occlusion, drastic illumination changes, and large pose variations. Experimental results show that CEST consistently outperforms state-of-the-art trackers.

In this paper, we formulate human action recognition as a novel Multi-Task Sparse Learning(MTSL) framework which aims to construct a test sample with multiple features from as few bases as possible. Learning the sparse representation under each feature modality is considered as a single task in MTSL. Since the tasks are generated from multiple features associated with the same visual input, they are not independent but inter-related. We introduce a Beta process(BP) prior to the hierarchical MTSL model, which efficiently learns a compact dictionary and infers the sparse structure shared across all the tasks. The MTSL model enforces the robustness in coefficient estimation compared with performing each task independently. Besides, the sparseness is achieved via the Beta process formulation rather than the computationally expensive L1 norm penalty. In terms of non-informative gamma hyper-priors, the sparsity level is totally decided by the data. Finally, the learning problem is solved by Gibbs sampling inference which estimates the full posterior on the model parameters. Experimental results on the KTH and UCF sports datasets demonstrate the effectiveness of the proposed MTSL approach for action recognition.

Multiple feature object representation has been proved as a robust approach for visual tracking. Different types of situations such as occlusion, rotation and illumination may occur during tracking, especially long sequences. Robust tracking could be obtained as multiple features could complement each other. In this paper, we cast visual tracking as a novel multi-task sparse learning problem and exploit various types of visual features, such as intensity, color, texture and edge, where each feature can be sparsely represented by a linear combination of atoms from an adaptive feature template. We use an on-line feature selection mechanism based on the two-class variance ratio measure, applied to log likelihood distributions computed with respect to a given feature from samples of object and background pixels. The proposed method is integrated in a particle filtering framework. We jointly consider the underlying relationship across different particles, and tackle it in a unified robust multi-task formulation. In addition, to capture the frequently emerging outlier tasks, we make fully use of a decomposition model which enables a more robust and accurate approximation. We show that the proposed model can be efficiently solved using the Alternating direction method of multipliers (ADMM) with a small number of closed-form updates. Four types of features are implemented and tested on numerous benchmark video sequences. Both the qualitative and quantitative results demonstrate the superior performance of the proposed approach compared to 9 state of-the-art trackers.

Dual feature correlation guided multi-task learning for Alzheimer's disease prediction

Several studies have demonstrated that dietary patterns identified by a posteriori and hybrid methods are associated with gastrointestinal (GI) cancer risk and mortality. These studies applied different methods for analyzing dietary data and reported inconsistent findings. This systematic review and meta-analysis were aimed to determine the association between dietary patterns, derived using principal component analysis (PCA) and reduced rank regression (RRR), and GI cancer risk and GI cancer-caused mortality. Articles published up to June 2023 in English were eligible for inclusion. The Medline, SCOPUS, Cochrane Library, CINHAL, PsycINFO, ProQuest, and Web of Sciences databases were used to identify prospective studies. The Preferred Reporting Item for Systematic Review and Meta-analysis Protocol 2020 was used to report results. A total of 28 studies were eligible for inclusion. Varied approaches to deriving dietary patterns were used, including PCA (n = 22), RRR (n = 2), combined PCA and RRR (n = 1), cluster analysis (CA; n = 2) and combined PCA and CA (n = 1). Two dietary patterns, "healthy" and "unhealthy," were derived using PCA and RRR. The healthy dietary pattern was characterized by a higher intake of fruits, whole grains, legumes, vegetables, milk, and other dairy products, whereas the unhealthy dietary pattern was characterized by a higher intake of red and processed meat, alcohol, and both refined and sugar-sweetened beverages. The findings indicated that the PCA-derived healthy dietary pattern was associated with an 8% reduced risk (relative risk [RR], 0.92; 95% CI, 0.87-0.98), and the unhealthy dietary pattern was associated with a 14% increased risk (RR, 1.14; 95% CI, 1.07-1.22) of GI cancers. Similarly, the RRR-derived healthy dietary pattern (RR, 0.83; 95% CI, 0.61-1.12) may be associated with reduced risk of GI cancers. In contrast, the RRR-derived unhealthy dietary pattern (RR, 0.93; 95% CI, 0.57-1.52) had no association with a reduced risk of GI cancers. Similarly, evidence suggested that PCA-derived healthy dietary patterns may reduce the risk of death from GI cancers, whereas PCA-derived unhealthy dietary patterns may increase the risk. Findings from prospective studies on the association of PCA-derived dietary patterns and the risk of GI cancers support the evidence of healthy and unhealthy dietary patterns as either protective or risk-increasing factors for GI cancers and for survivorship, respectively. The findings also suggest that the RRR-derived healthy dietary pattern reduces the risk of GI cancers (albeit with low precision), but no association was found for the RRR-derived unhealthy dietary pattern. Prospective studies are required to further clarify disparities in the association between PCA- and RRR-derived dietary patterns and the risk of GI cancers. Systematic review registration: PROSPERO registration no. CRD42022321644.

Image fusion is one of the important techniques for image information enhancing. In order to utilize respective information from different remote sensing images, we propose a new image fusion method based on the Principal Component Analysis (PCA) and feature product of wavelet transform. Firstly, the multi-spectral image is transformed with PCA. Secondly, the histogram-matched panchromatic image and the first principal component are decomposed into wavelet coefficients respectively. Thirdly, the first principal component of the multi-spectral image and the panchromatic image are merged with feature product of wavelet based fusion method, and the former is replaced with the merged data. Finally, the new multi-spectral image is obtained by inverse PCA. Some evaluation parameters are suggested and applied to compare the new method with those of PCA method, the combined PCA and traditional wavelet method and the combined PCA and local deviation of wavelet method. Subjective visual effect and objective statistical results indicate that the performance of the new method is better than those methods. It not only preserves spectral information of the original multi-spectral image well, but also enhances spatial detail information greatly.

The alternating direction method of multipliers (ADMM) is an algorithm for solving large-scale data optimization problems in machine learning. In order to reduce the communication delay in a distributed environment, asynchronous distributed ADMM (AD-ADMM) was proposed. However, due to the unbalance process arrival pattern existing in the multiprocessor cluster, the communication of the star structure used in AD-ADMM is inefficient. Moreover, the load in the entire cluster is unbalanced, resulting in a decrease of the data processing capacity. This paper proposes a hierarchical parameter server communication structure (HPS) and an asynchronous distributed ADMM (HAD-ADMM). The algorithm mitigates the unbalanced arrival problem through process grouping and scattered updating global variable, which basically achieves load balancing. Experiments show that the HAD-ADMM is highly efficient in a large-scale distributed environment and has no significant impact on convergence.

The hypothesis that elderly individuals are less likely than young adults to connect target and contextual information was tested. Young and elderly adults were presented with a number of slides, each of which contained a word superimposed in the center of a background picture of a landscape or cityscape. Half of the subjects were told to remember the words and half were told to remember the word-and-background pairs. All subjects were then tested for their recognition memory of the word-and-background pairs. The results indicate that elderly adults have greater difficulty than young adults remembering the connections between words and background pictures but that this occurs whether the pictures are target information or contextual information. Therefore, the results of this study provide no support for the notion that elderly adults have a specific contextual encoding deficit.

Aiming at solving large-scale learning problems, this paper studies distributed optimization methods based on the alternating direction method of multipliers (ADMM). By formulating the learning problem as a consensus problem, the ADMM can be used to solve the consensus problem in a fully parallel fashion over a computer network with a star topology. However, traditional synchronized computation does not scale well with the problem size, as the speed of the algorithm is limited by the slowest workers. This is particularly true in a heterogeneous network where the computing nodes experience different computation and communication delays. In this paper, we propose an asynchronous distributed ADMM (AD-AMM) which can effectively improve the time efficiency of distributed optimization. Our main interest lies in analyzing the convergence conditions of the AD-ADMM, under the popular partially asynchronous model, which is defined based on a maximum tolerable delay of the network. Specifically, by considering general and possibly non-convex cost functions, we show that the AD-ADMM is guaranteed to converge to the set of Karush-Kuhn-Tucker (KKT) points as long as the algorithm parameters are chosen appropriately according to the network delay. We further illustrate that the asynchrony of the ADMM has to be handled with care, as slightly modifying the implementation of the AD-ADMM can jeopardize the algorithm convergence, even under a standard convex setting.

Exploiting context information in a web search engine helps fine-tuning web services and applications to deliver custom-made information to end users. While context, including user and environment information, cannot be exploited efficiently in the wired Internet interaction type, it is becoming accessible with the mobile web where users have an intimate relationship with their handsets. In this type of interaction, context plays a significant role enhancing information search and therefore, allowing a search engine to detect relevant content in all digital forms and formats. This chapter proposes a context model and an architecture that promote integration of context information for individuals and social communities to add value to their interaction with the mobile web. The architecture relies on efficient knowledge management of multimedia resources for a wide range of applications and web services. The research is illustrated with a corporate case study showing how efficient context integration improves usability of a mobile search engine.

We tackle the problem of multi-task learningMulti-Task Learning (MTL) (MTL) for solving the correlated visual dense prediction tasks using monocular image source. With further restrictions on the model parameters, the soft-shared MTL Convnets (CNN) feature distinct models for each individual task. Hard-sharingHard-sharing-based models, on the other hand, have shared encoders but individual decoders for every task. MTLMulti-Task Learning (MTL) models have demonstrated satisfactory performances for pixel-wise dense prediction tasks like semantic segmentation, surface-normal estimation, and depth estimation from the monocular inputs. In general, they impose two inherent drawbacks: (1) such models have constraints with no leverage on the inter-task knowledge, which hinders the performance boost while jointly trained, and (2) explicitly optimization of each task-specific network is required. Incorporating the abovementioned issues into soft and hard-sharingHard-sharing-based MTL models can immensely enhance MTLMulti-Task Learning (MTL) performance. To that end, we present SD-MTCNN, hard-sharing-based and S3DMT-Net, soft-sharingSoft-sharing-based novel MTL networks. Here, we follow tries to inherit characteristics from deeper CNNConvolutional Neural Network (CNN) layers/feature maps into shallower CNN layers, hence helps in increasing the network's bandwidth. We also utilize the notion of sharing the features of the task-specific encoders, where the task-specific encoder’s feature-maps are communicated amongst each other, aka as cross-task interactions. We examine our self-distilled MTLsMulti-Task Learning (MTL) performance on two different types of visual scenes: Urban (CityScapes and ISPRS) and indoor (NYUv2 and Mini-Taskonomy), and we observe notable gains in all tasks as compared to the referred methodologies.

Forensic feature extraction of document paper using periodic marks: PCA and t-SNE for manufacturer discrimination and document dating.

Previous research has demonstrated that patients with aphasia benefit from context that occurs prior to target information, particularly if it predicts the forthcoming information. However, contextual information frequently occurs concomitantly with or following target information. This study assesses aphasic patients' ability to benefit from context that occurs before or after target information. The results demonstrate that both contextual conditions generate significantly better comprehension than is achieved by presenting the target information by itself. Simply repeating the target information once does not generate significantly better comprehension. These results indicate that patients with aphasia can integrate both target and contextual information, regardless of order, before responding.

Because of the powerful computing and storage capability in cloud computing, machine learning as a service (MLaaS) has recently been valued by the organizations for machine learning training over some related representative datasets. When these datasets are collected from different organizations and have different distributions, multi-task learning (MTL) is usually used to improve the generalization performance by scheduling the related training tasks into the virtual machines in MLaaS and transferring the related knowledge between those tasks. However, because of concerns about privacy breaches (e.g., property inference attack and model inverse attack), organizations cannot directly outsource their training data to MLaaS or share their extracted knowledge in plaintext, especially the organizations in sensitive domains. In this article, we propose a novel privacy-preserving mechanism for distributed MTL, namely NOInfer, to allow several task nodes to train the model locally and transfer their shared knowledge privately. Specifically, we construct a single-server architecture to achieve the private MTL, which protects task nodes’ local data even if n-1 out of n nodes colluded. Then, a new protocol for the Alternating Direction Method of Multipliers (ADMM) is designed to perform the privacy-preserving model training, which resists the inference attack through the intermediate results and ensures that the training efficiency is independent of the number of training samples. When releasing the trained model, we also design a differentially private model releasing mechanism to resist the membership inference attack. Furthermore, we analyze the privacy preservation and efficiency of NOInfer in theory. Finally, we evaluate our NOInfer over two testing datasets and evaluation results demonstrate that NOInfer efficiently and effectively achieves the distributed MTL.