Xiamen University

About: Xiamen University is a education organization based out in Amoy, Fujian, China. It is known for research contribution in the topics: Catalysis & Population. The organization has 50472 authors who have published 54480 publications receiving 1058239 citations. The organization is also known as: Amoy University & Xiàmén Dàxué.

HRank: Filter Pruning Using High-Rank Feature Map

Abstract: Neural network pruning offers a promising prospect to facilitate deploying deep neural networks on resource-limited devices. However, existing methods are still challenged by the training inefficiency and labor cost in pruning designs, due to missing theoretical guidance of non-salient network components. In this paper, we propose a novel filter pruning method by exploring the High Rank of feature maps (HRank). Our HRank is inspired by the discovery that the average rank of multiple feature maps generated by a single filter is always the same, regardless of the number of image batches CNNs receive. Based on HRank, we develop a method that is mathematically formulated to prune filters with low-rank feature maps. The principle behind our pruning is that low-rank feature maps contain less information, and thus pruned results can be easily reproduced. Besides, we experimentally show that weights with high-rank feature maps contain more important information, such that even when a portion is not updated, very little damage would be done to the model performance. Without introducing any additional constraints, HRank leads to significant improvements over the state-of-the-arts in terms of FLOPs and parameters reduction, with similar accuracies. For example, with ResNet-110, we achieve a 58.2%-FLOPs reduction by removing 59.2% of the parameters, with only a small loss of $0.14\%$ in top-1 accuracy on CIFAR-10. With Res-50, we achieve a 43.8%-FLOPs reduction by removing 36.7% of the parameters, with only a loss of 1.17% in the top-1 accuracy on ImageNet. The codes can be available at https://github.com/lmbxmu/HRank.

Chemistry with Electrochemically Generated N-Centered Radicals

Abstract: N-centered radicals are versatile reaction intermediates that can react with various π systems to construct C-N bonds. Current methods for generating N-centered radicals usually involve the cleavage of an N-heteroatom bond; however, similar strategies that are applicable to N-H bonds prove to be more challenging to develop and therefore are attracting increasing attention. In this Account, we summarize our recent efforts in the development of electrochemical methods for the generation and synthetic utilization of N-centered radicals. In our studies, N-aryl amidyl radical, amidinyl radical and iminyl radical cation intermediates are generated from N-H precursors through direct electrolysis or indirect electrolysis assisted by a redox catalyst. In addition, an electrocatalytic method that converts oximes to iminoxyl radicals has also been developed. The electrophilic amidyl radical intermediates can participate in 5-exo or 6-exo cyclization with alkenes and alkynes to afford C-centered radicals, which can then undergo various transformations such as H atom abstraction, single-electron transfer oxidation to a carbocation, cyclization, or aromatic substitution, leading to a diverse range of N-heterocyclic products. Furthermore, amidinyl radicals, iminyl radical cations, and iminoxyl radicals can undergo intramolecular aromatic substitution to afford various N-heteroaromatic compounds. Importantly, the electrochemical reaction can be channeled toward a specific product despite the presence of other competing pathways. For a successful electrosynthesis, it is important to take into consideration of both the electron transfer steps associated with the electrode and the nonelectrode related processes. A unique feature of electrochemistry is the simultaneous occurrence of anodic oxidation and cathodic reduction, which, as this Account demonstrates, allows the dehydrogenative transformations to proceed through H2 evolution without the need for chemical oxidants. In addition, cathodic solvent reduction can continuously generate a low concentration of base, which facilitates anodic substrate oxidation. Such a mechanistic paradigm obviates the need for stoichiometric strong bases and avoids base-promoted decomposition of sensitive substrates or products. Furthermore, electrode materials can also be adjusted to control the reaction outcome, as demonstrated by the synthesis of N-heteroaromatics and the corresponding N-oxides from biaryl ketoximes.

Exploiting the Reversible Covalent Bonding of Boronic Acids: Recognition, Sensing, and Assembly

Abstract: Boronic acids can interact with Lewis bases to generate boronate anions, and they can also bind with diol units to form cyclic boronate esters. Boronic acid based receptor designs originated when L...

RGBD Salient Object Detection: A Benchmark and Algorithms

Abstract: Although depth information plays an important role in the human vision system, it is not yet well-explored in existing visual saliency computational models. In this work, we first introduce a large scale RGBD image dataset to address the problem of data deficiency in current research of RGBD salient object detection. To make sure that most existing RGB saliency models can still be adequate in RGBD scenarios, we continue to provide a simple fusion framework that combines existing RGB-produced saliency with new depth-induced saliency, the former one is estimated from existing RGB models while the latter one is based on the proposed multi-contextual contrast model. Moreover, a specialized multi-stage RGBD model is also proposed which takes account of both depth and appearance cues derived from low-level feature contrast, mid-level region grouping and high-level priors enhancement. Extensive experiments show the effectiveness and superiority of our model which can accurately locate the salient objects from RGBD images, and also assign consistent saliency values for the target objects.