Auto-Encoder (AE)-based deep subspace clustering (DSC) methods have achieved impressive performance due to the powerful representation extracted using deep neural networks while prioritizing ...categorical separability. However, self-reconstruction loss of an AE ignores rich useful relation information and might lead to indiscriminative representation, which inevitably degrades the clustering performance. It is also challenging to learn high-level similarity without feeding semantic labels. Another unsolved problem facing DSC is the huge memory cost due to <inline-formula> <tex-math notation="LaTeX">n\times n </tex-math></inline-formula> similarity matrix, which is incurred by the self-expression layer between an encoder and decoder. To tackle these problems, we use pairwise similarity to weigh the reconstruction loss to capture local structure information, while a similarity is learned by the self-expression layer. Pseudo-graphs and pseudo-labels, which allow benefiting from uncertain knowledge acquired during network training, are further employed to supervise similarity learning. Joint learning and iterative training facilitate to obtain an overall optimal solution. Extensive experiments on benchmark datasets demonstrate the superiority of our approach. By combining with the <inline-formula> <tex-math notation="LaTeX">k </tex-math></inline-formula>-nearest neighbors algorithm, we further show that our method can address the large-scale and out-of-sample problems. The source code of our method is available: https://github.com/sckangz/SelfsupervisedSC .
Constructing the adjacency graph is fundamental to graph-based clustering. Graph learning in kernel space has shown impressive performance on a number of benchmark data sets. However, its performance ...is largely determined by the chosen kernel matrix. To address this issue, previous multiple kernel learning algorithm has been applied to learn an optimal kernel from a group of predefined kernels. This approach might be sensitive to noise and limits the representation ability of the consensus kernel. In contrast to existing methods, we propose to learn a low-rank kernel matrix which exploits the similarity nature of the kernel matrix and seeks an optimal kernel from the neighborhood of candidate kernels. By formulating graph construction and kernel learning in a unified framework, the graph and consensus kernel can be iteratively enhanced by each other. Extensive experimental results validate the efficacy of the proposed method.
The power conditioning interface circuit plays a crucial role in the energy harvesting (EH) capability enhancement in piezoelectric EH (PEH) systems. The principle of most existing synchronized ...switch circuits for PEH enhancement was summarized under the recently introduced synchronized multiple bias-flip (SMBF) model. The optimal bias-flip strategy derived from SMBF provides valuable insights for future circuit development. Based on such theoretical foundation, this paper implements a new interface circuit called the parallel synchronized triple bias-flip (P-S3BF) for further boosting the harvesting capability beyond the state-of-the-art solutions. By sophisticatedly allocating more bias-flip actions, it makes the best compromise between higher extracted power and lower dissipated power for power conditioning, toward maximum net harvested power. Moreover, the bias voltage reference in P-S3BF is self-contained and self-adaptive. Compared to the other existing solutions involving active voltage manipulations, these two conveniences brought in by P-S3BF are the most significant and necessary features, which facilitate the practical stand-alone implementation of P-S3BF under different periodic vibrations. Experimental result shows that, with the piezoelectric cantilever in use, which vibrates at a constant deflection magnitude, the P-S3BF circuit harvests 24.5% more power than its single bias-flip counterpart, the extensively studied parallel synchronized switch harvesting on inductor; and 287.6% more power than its null bias-flip counterpart, the standard EH bridge rectifier.
Abstract
K‐metal batteries have become one of the promising candidates for the large‐scale energy storage owing to the virtually inexhaustible and widely potassium resources. The uneven K
+
...deposition and dendrite growth on the anode causes the batteries prematurely failure to limit the further application. An integrated K‐metal anode is constructed by cold‐rolling K metal with a potassiphilic porous interconnected mediator. Based on the experimental results and theoretical calculations, it demonstrates that the potassiphilic porous interconnected mediator boosts the mass transportation of K‐metal anode by the K affinity enhancement, which decreases the concentration polarization and makes a dendrite‐free K‐metal anode interface. The interconnected porous structure mitigates the internal stress generated during repetitive deposition/stripping, enabling minimized the generation of electrode collapse. As a result, a durable K‐metal anode with excellent cycling ability of exceed 1, 000 h at 1 mA cm
−2
/1 mAh cm
−2
and lower polarization voltage in carbonate electrolyte is obtained. This proposed integrated anode with fast K
+
kinetics fabricated by a repeated cold rolling and folding process provides a new avenue for constructing a high‐performance dendrites‐free anode for K‐metal batteries.
•A graph learning framework, which captures both the global and local structure in data, is proposed.•Theoretical analysis builds the connections of our model to k-means, spectral clustering, and ...kernel k-means.•Extensions to semi-supervised classification and multiple kernel learning are presented.
Graphs have become increasingly popular in modeling structures and interactions in a wide variety of problems during the last decade. Graph-based clustering and semi-supervised classification techniques have shown impressive performance. This paper proposes a graph learning framework to preserve both the local and global structure of data. Specifically, our method uses the self-expressiveness of samples to capture the global structure and adaptive neighbor approach to respect the local structure. Furthermore, most existing graph-based methods conduct clustering and semi-supervised classification on the graph learned from the original data matrix, which doesn’t have explicit cluster structure, thus they might not achieve the optimal performance. By considering rank constraint, the achieved graph will have exactly c connected components if there are c clusters or classes. As a byproduct of this, graph learning and label inference are jointly and iteratively implemented in a principled way. Theoretically, we show that our model is equivalent to a combination of kernel k-means and k-means methods under certain condition. Extensive experiments on clustering and semi-supervised classification demonstrate that the proposed method outperforms other state-of-the-art methods.
PurposeSome studies have claimed that Chinese thinker Hu Shi (or Hu Shih) received and responded to John Dewey's educational ideas only at a theoretical level and did little for education at a ...practical level. This paper reexamines Hu's reception of Dewey's ideas with a focus on how he used those ideas to solve China's educational and social problems during the late 1910s and 1920s.Design/methodology/approachThis paper draws upon what Schriewer (2012) has called “theories of reception.” Rather than focusing on the international dissemination of ideas and knowledge, this approach emphasizes the reception of foreign ideas from the perspective and needs of the receiver, interpreter and/or reader who apprehends such ideas within a particular socio–cultural context.FindingsThis paper finds that Hu not only received — and examined — Dewey's educational ideas in a systematic way, but also used them pragmatically to reform China's systems of education as part of the New Culture Movement after 1919.Originality/valueThis research offers a new understanding of Hu's reception of Dewey's educational ideas. It shows that Hu was not merely a “thinker” in the field of education but also a “doer” who sought to apply Dewey's ideas in practice. This new view allows us to reevaluate Hu's role in the modernization of Chinese education.
In order to control the adverse effect of vehicles overloading infrastructure and traffic safety, weight-in-motion (WIM)-related research has drawn growing attention. To address the high cost of ...current piezoelectric sensors in installation and maintenance, a study on developing a low-cost piezoceramic sensing system is presented in this paper. The proposed system features distributed monitoring and integrated packaging, for calculating vehicle’s dynamic load and its wheel position. Results from the laboratory tests show that the total output of the sensing system increases linearly with the increase of the peak load when the loading amplitude is 5–25 kN (equivalent to the half-axis load of 20–100 kN); when the loading frequency is between 15 Hz and 19 Hz (equivalent to a speed of 17.8–23.2 km/h), the total output of the system fluctuates around a value of 1.305 V. Combined with finite-element simulation, the system can locate load lateral position with a resolution of 120 mm. Due to the protection packaging, the peak load transferred to the sensing units is approximately 4.36% of the applied peak load. The study indicates the proposed system can provide a promising low-cost, reliable and practical alternative for current WIM systems.
Multi-view subspace clustering (MVSC) has drawn significant attention in recent study. In this paper, we propose a novel approach to MVSC. First, the new method is capable of preserving high-order ...neighbor information of the data, which provides essential and complicated underlying relationships of the data that is not straightforwardly preserved by the first-order neighbors. Second, we design log-based nonconvex approximations to both tensor rank and tensor sparsity, which are effective and more accurate than the convex approximations. For the associated shrinkage problems, we provide elegant theoretical results for the closed-form solutions, for which the convergence is guaranteed by theoretical analysis. Moreover, the new approximations have some interesting properties of shrinkage effects, which are guaranteed by elegant theoretical results. Extensive experimental results confirm the effectiveness of the proposed method.
We propose a new subspace clustering method that integrates feature selection into subspace clustering. Rather than using all features to construct a low-rank representation of the data, we find such ...a representation using only relevant features, which helps in revealing more accurate data relationships. Two variants are proposed by using both convex and nonconvex rank approximations. Extensive experimental results confirm the effectiveness of the proposed method and models.
Purpose:
This article aims to conduct a critical analysis of key-competences-based education in China today by revealing three main challenges that Chinese educational practitioners, researchers, and ...policymakers might face.
Design/Approach/Methods:
Against the backdrop of international and Chinese policy and the practice of key-competences-based education, this article focuses on a critical discourse analysis of this educational movement in China.
Findings:
China places great emphasis on the personal, cultural, and moral dimensions in generating a framework of competences, paying considerable attention to the well-being of children; hence it currently endeavors to provide a consistent educational framework to educate the young for the all-round development. However, three challenges are salient: the dilemma between Core-Suyang (Core-Competence) and the education for all-round development of the young, the limitations of design of “subject core-competences,” and a lack of consideration of subjectification as purpose of education.
Originality/Value:
Education which merely focuses on a list of key-competences might not only exclude non-key-competences but also exclude something unforeseen and unknown in the process of educational adventure. It is therefore suggested that a holistic education needs to be engaged in with an encompassing and dynamic view of purpose(s) of education, in which the education for subjectification is at stake in our educational endeavor.