Significant progress has been made in recent years in image captioning, an active topic in the fields of vision and language. However, existing methods tend to yield overly general captions and consist of some of the most frequent words/phrases, resulting in inaccurate and indistinguishable descriptions (see Figure 1). This is primarily due to (i) the conservative characteristic of traditional training objectives that drives the model to generate correct but hardly discriminative captions for similar images and (ii) the uneven word distribution of the ground-truth captions, which encourages generating highly frequent words/phrases while suppressing the less frequent but more concrete ones. In this work, we propose a novel global-local discriminative objective that is formulated on top of a reference model to facilitate generating fine-grained descriptive captions. Specifically, from a global perspective, we design a novel global discriminative constraint that pulls the generated sentence to better discern the corresponding image from all others in the entire dataset. From the local perspective, a local discriminative constraint is proposed to increase attention such that it emphasizes the less frequent but more concrete words/phrases, thus facilitating the generation of captions that better describe the visual details of the given images. We evaluate the proposed method on the widely used MS-COCO dataset, where it outperforms the baseline methods by a sizable margin and achieves competitive performance over existing leading approaches. We also conduct self-retrieval experiments to demonstrate the discriminability of the proposed method.
10 November 2017 in IEEE Transactions on Knowledge and Data Engineering
In this paper, we propose a prototype-based classification model for evolving data streams, called SyncStream, which allows dynamically modeling time-changing concepts, making predictions in a local fashion. Instead of learning a single model on a fixed or adaptive sliding window of historical data or ensemble learning a set of weighted base classifiers, SyncStream captures evolving concepts by dynamically maintaining a set of prototypes in a proposed P-Tree, which are obtained based on the error-driven representativeness learning and synchronization-inspired constrained clustering. To identify abrupt concept drifts in data streams, PCA and statistical analysis based heuristic approaches have been introduced. To further learn the associations among distributed data streams, the extended P-Tree structure and KNN-style strategy are introduced. We demonstrate that our new data stream classification approach has several attractive benefits: (a) SyncStream is capable of dynamically modeling the evolving concepts from even a small set of prototypes. (b) Owing to synchronization-based constrained clustering and P-Tree, SyncStream supports efficient and effective data representation and maintenance. (c) SyncStream is also tolerant of inappropriate or noisy examples via error-driven representativeness learning. (d) SyncStream allows learning relationship among distributed data streams at the instance level. The experimental results indicate its efficiency and effectiveness.
Sensor drift is a common issue in E-Nose systems and various drift compensation methods have received fruitful results in recent years. Although the accuracy for recognizing diverse gases under drift conditions has been largely enhanced, few of these methods considered online processing scenarios. In this paper, we focus on building online drift compensation model by transforming two domain adaptation based methods into their online learning versions, which allow the recognition models to adapt to the changes of sensor responses in a time-efficient manner without losing the high accuracy. Experimental results using three different settings confirm that the proposed methods save large processing time when compared with their offline versions, and outperform other drift compensation methods in recognition accuracy.
5 February 2018 in IEEE Transactions on Network Science and Engineering
Due to the tremendous growth of online social networks in both participants and collected contents, social data publication has provided an opportunity for numerous services. However, neglectfully publishing all the contents leads to severe disclosure of sensitive information due to diverse user behaviors. Therefore, there should be a thoroughly designed framework for data publication in online social networks that considers users heterogeneous privacy preferences and the correlations among participants. This work proposes a novel mechanism for data publication that achieves high performance while preserving privacy and guaranteeing fairness among users. To derive the optimal scheme for data publication is NP-complete. Thus we propose a heuristic algorithm to determine the contents to be published which takes advantage of the sets of sensitive contents for each user and the correlation among them. The theoretical analysis proves the effectiveness and feasibility of the mechanism. The evaluations towards a real-world dataset reveal that the proposed algorithm outperforms the existing results.
1 October 2020 in IEEE Transactions on Network Science and Engineering
Following the application of Deep Learning to graphic data, Graph Neural Networks (GNNs) have become the dominant method for Node Classification on graphs in recent years. To assign nodes with preset labels, most GNNs inherit the end-to-end way of Deep Learning in which node features are input to models while labels of pre-classified nodes are used for supervised learning. However, while these methods can make full use of node features and their associations, they treat labels separately and ignore the structural information of those labels. To utilize information on label structures, this paper proposes a method called 3ference that infers from references with differences. Specifically, 3ference predicts what label a node has according to the features of that node in concatenation with both features and labels of its relevant nodes. With the additional information on labels of relevant nodes, 3ference captures the transition pattern of labels between nodes, as subsequent analysis and visualization revealed. Experiments on a synthetic graph and seven real-world graphs proved that this knowledge about label associations helps 3ference to predict accurately with fewer parameters, fewer pre-classified nodes, and varying label patterns compared with GNNs.