Yuefan Deng - Academia.edu (original) (raw)

Papers by Yuefan Deng

Frontiers in signal processing, Jan 31, 2021

Information aggregation is an essential component of text encoding, but it has been paid less att... more Information aggregation is an essential component of text encoding, but it has been paid less attention. The pooling-based (max or average pooling) aggregation method is a bottom-up and passive aggregation method, and loses a lot of important information. Recently, attention mechanism and dynamic routing policy are separately used to aggregate information, but their aggregation capabilities can be further improved. In this paper, we proposed an novel aggregation method combining attention mechanism and dynamic routing, which can strengthen the ability of information aggregation and improve the quality of text encoding. Then, a novel Leaky Natural Logarithm (LNL) squash function is designed to alleviate the "saturation" problem of the squash function of the original dynamic routing. Layer Normalization is added to the dynamic routing policy for speeding up routing convergence as well. A series of experiments are conducted on five text classification benchmarks. Experimental results show that our method outperforms other aggregating methods.

Aiming at the problem that the Quantum-behaved Pigeon-inspired optimization Algorithm is easy to ... more Aiming at the problem that the Quantum-behaved Pigeon-inspired optimization Algorithm is easy to fall into the local optimum and is lack of fine search ability, a new algorithm based on particle-best mutation is proposed. Firstly, aiming at the map and compass operator characteristics of solve rough search, Cauthy mutation based on particle-best was adopted for the pigeons when they have not evolved at this stage, so as to enhance the global search ability. Then, aiming at the landmark operator characteristics of solve fine search, Gaussian mutation after updating by landmark operator was adopted for the pigeons when they have not evolved at this stage, so as to enhance the local search ability. Finally, model of pigeons’ number factor which likes map and compass factor was built to keep the pigeons in the landmark operator at a certain amount, so as to get the optimal solution. The function test results show that the search speed and accuracy of the improved algorithm are better than other popular algorithms.

IACR Cryptology ePrint Archive, 2016

Attribute-Based Encryption (ABE) is a special type of public key encryption that allows users to ... more Attribute-Based Encryption (ABE) is a special type of public key encryption that allows users to share sensitive data efficiently through fine-grained access control. The security involved in existing ABE systems is currently insufficient. These systems are usually built on the Decisional Bilinear Diffie-Hellman (DBDH) assumption or the q-type DBDH assumption, which is stronger than the DBDH assumption. However, once the DBDH assumption is unsecure, all concerned ABEs become vulnerable to threats. To address this problem, the k-BDH assumption family proposed by Benson et al. is adopted. Any assumption in the k-BDH assumption family is associated with parameter k and becomes strictly weaker as k increased. We propose a framework to implement Ciphertext-Policy Attribute Based Encryption (CP-ABE) under the arbitrary assumption in the k-BDH assumption family. When the k-BDH assumption in the k-BDH assumption family becomes unsecure, where k-BDH is the assumption on which our ABE relies, the scheme can be shifted to rely on the l-BDH assumption instead, where l > k. This condition guarantees security as the underlying assumption of our scheme becomes weaker. In addition, we define the formal security model of our schemes and prove the security of CP-ABE in the selective attribute model.

Computer Engineering, 2009

【Abstract】Aiming at the problem that current database could not predict pages will be access, thi... more 【Abstract】Aiming at the problem that current database could not predict pages will be access, this paper presents application Call Graph(CG) model and correlative CG algorithms, and the performance of CG algorithm is analyzed. In the experiments, basing on GKD-Base, this paper realizes CG algorithms, and predicts the pages which will not be accessed in a period of time, then according to the prediction, these pages could be replaced out of page buffer ahead of schedule to optimize buffer space. Experimental results show that the database performance does not reduce after adding CG algorithms in database kernel, and instead, due to buffer space optimization, SQL execution speed is improved.

SiPS ... design and implementation, Oct 1, 2007

... B. Parity Calculationfor SubByte Module Main computations in MixColumn moduleare xtime and Si... more ... B. Parity Calculationfor SubByte Module Main computations in MixColumn moduleare xtime and Since SubByte operation is the only nonlinear module in XOR operation. ... E. Parity Calculationfor Key Expansion Module [xtime(So0j)+ S0O.]+ Sjj + S2,j+ xtime(S3,j) ...

This paper proposes a new verifiable visual cryptography scheme for general access structures usi... more This paper proposes a new verifiable visual cryptography scheme for general access structures using pisigma neural networks (VVCSPSN), which is based on probabilistic signature scheme (PSS), which is considered as security and effective verification method. Compared to other high-order networks, PSN has a highly regular structure, needs a much smaller number of weights and less training time. Using PSN's capability of large-scale parallel classification, VCSPSN reduces the information communication rate greatly, makes best known upper bound polynomial, and distinguishes the deferent information in secret image.

DEStech Transactions on Computer Science and Engineering, May 5, 2017

One drawback of the traditional KP-ABE is the limitation for expressing the hierarchical structur... more One drawback of the traditional KP-ABE is the limitation for expressing the hierarchical structure of attributes. In this paper, we firstly propose a fully secure Key-Policy Hierarchical Attribute-based encryption (KP-HABE). In KP-HABE, attribute is expressed by a vector in order to facilitate the indication of the hierarchical structure of attribute. The technique of composite order bilinear pairing and linear secret sharing scheme is used to build the KP-HABE scheme.

IS&T International Symposium on Electronic Imaging Science and Technology, Jan 29, 2017

For the virtual reality (VR) applications, it is important to quantify the just-noticeable-differ... more For the virtual reality (VR) applications, it is important to quantify the just-noticeable-difference (JND) profile of the image such that immersive experience could be provided. In this paper, we first propose a test plan of subjective quality assessment for VR image, then a foveated JND (FJND) model that is capable of modeling the basic vision properties to quantify the correlation between eccentricity and conventional JND. The psychophysical experiments discover the relationship among the relevant factors and quantify the foveated JND that could be used for immersive VR applications.

In this paper, we present an optimal frame synchronization scheme for DVB-S2 system. The scheme i... more In this paper, we present an optimal frame synchronization scheme for DVB-S2 system. The scheme is based on differential detection and non-threshold peak search algorithm to achieve frame synchronization. By further analyzing the structure of the Physical Layer Header (PLHEADER) that consists of Start of Frame (SOF) and Physical Layer Signalling Code (PLSC), we find some useful properties that can enhance the frame synchronization and reduce the complexity of its implementation. We also introduce some optimizations into the algorithm. Especially, the modifications made on the peak search algorithm improve the overall performance of the frame synchronization. The experimental results show that the optimal frame synchronization can be achieved under large carrier frequency offset (as large as 5 MHz at 25 MSps) and low SNR threshold of - 2.3 dB. Compared with previously proposed approaches, the scheme described in this paper takes less time to acquire frame synchronization and seems to be more robust.

Information Sciences, May 1, 2020

With the aggravating trend of aging population in modern societies, Mobile Healthcare Social Netw... more With the aggravating trend of aging population in modern societies, Mobile Healthcare Social Network (MHSN) is an attracting platform for improving the life quality of elderly patients and offering mutual supports. In a sustainable MHSN system, one essential issue is to protect sensitive healthcare information while providing information-sharing for patients with multiple symptoms. To achieve multi-symptom matching and cross-domain communications for patients in MHSNs, we propose a new Secret Handshake scheme with Multi-Symptom intersection (MSSH), which is derived from a linear-complexity Authorized Private Set Intersection (APSI) based on Schnorr signature. In MSSH , two anonymous patients, who have registered with independent healthcare centers, are able to successfully execute an affiliation-hiding mutual anonymous authentication only if their target authentication policies are satisfied. The authentication policy demands that the designated healthcare center should be matched and the cardinality of their symptoms intersection is not less than a threshold value. Meanwhile, the symptoms outside of the set intersection remain confidential. The formal analysis proves MSSH is secure under the Random Oracle Model (ROM). The experimental results demonstrate that MSSH is efficient and feasible in MHSNs.

arXiv (Cornell University), Aug 12, 2021

While deep neural networks have shown impressive performance in many tasks, they are fragile to c... more While deep neural networks have shown impressive performance in many tasks, they are fragile to carefully designed adversarial attacks. We propose a novel adversarial training-based model by Attention Guided Knowledge Distillation and Bi-directional Metric Learning (AGKD-BML). The attention knowledge is obtained from a weight-fixed model trained on a clean dataset, referred to as a teacher model, and transferred to a model that is under training on adversarial examples (AEs), referred to as a student model. In this way, the student model is able to focus on the correct region, as well as correcting the intermediate features corrupted by AEs to eventually improve the model accuracy. Moreover, to efficiently regularize the representation in feature space, we propose a bidirectional metric learning. Specifically, given a clean image, it is first attacked to its most confusing class to get the forward AE. A clean image in the most confusing class is then randomly picked and attacked back to the original class to get the backward AE. A triplet loss is then used to shorten the representation distance between original image and its AE, while enlarge that between the forward and backward AEs. We conduct extensive adversarial robustness experiments on two widely used datasets with different attacks. Our proposed AGKD-BML model consistently outperforms the state-of-the-art approaches. The code of AGKD-BML will be available at: https://github.com/hongw579/AGKD-BML.

Aiming at the shortage of the fine search ability in the Quantum-behaved Pigeon-inspired Optimiza... more Aiming at the shortage of the fine search ability in the Quantum-behaved Pigeon-inspired Optimization Algorithm, a new algorithm with mutation disturbance is proposed. Firstly, aiming at the map and compass operator characteristics of solve rough search, Cauchy mutation is adopted for the pigeons when they have not evolved at this stage, so as to enhance the global search ability. Then, aiming at the landmark operator characteristics of solve fine search, Gaussian mutation is adopted for the pigeons when they have not evolved at this stage, so as to enhance the local search ability. Finally, model of pigeons' number factor which likes map and compass factor is built to keep the pigeons in the landmark operator at a certain amount, so as to get the optimal solution. The function test results show that the search speed and accuracy of the improved algorithm are better than other popular algorithms.

Aiming at the existing problem of single buffer replacement policy and the inability of predictin... more Aiming at the existing problem of single buffer replacement policy and the inability of predicting future SQLs execution in current database buffer replacement algorithm, proposed a new database buffer replacement algorithm---CG-ABR (call graph optimizing adaptive buffer replacement algorithm). CG-ABR could adapt to current SQL&#39;s access behavior automatically by getting it through query optimizer, and this algorithm optimized the management method

Springer eBooks, 2009

Chip Multi-Processor(CMP) allows multiple threads to execute simultaneously. Because threads shar... more Chip Multi-Processor(CMP) allows multiple threads to execute simultaneously. Because threads share various resources of CMP, such as L2-Cache, CMP system is inherently different from multiprocessors system and, CMP is also different from simultaneous multithreading (SMT). It could support more than two threads to execute simultaneously, and some executing units are owned by each core. We present hash join optimization based on shared cache CMP. Firstly, we propose multithreaded hash join execution framework based on Radix-Join algorithm, then we analyze the factors which affect performance of multithreaded Radix-Join algorithm in CMP. Basing on this analysis, we optimize the performance of various threads and their shared-cache access performance in the framework, and then theoretic analysis of speedup in multithreaded cluster partition phase is presents which could give some advices to cluster partition thread optimization. All of our algorithms are implemented in EaseDB. In the experiments, we evaluate performance of the multithreaded hash join execution framework, and the results show that our algorithm could effectively resolve cache access conflict and load imbalance in multithreaded environment. Hash join performance is improved.

Journal of Networks, Oct 1, 2011

In 1998, Blaze, Bleumer, and Strauss (BBS) proposed proxy re-signatures, in which a semi trusted ... more In 1998, Blaze, Bleumer, and Strauss (BBS) proposed proxy re-signatures, in which a semi trusted proxy acts as a translator between Alice and Bob. The proxy resignature schemes can be applied in many applications. However, the existing proxy re-signature schemes were all based on Diffie-Hellman assumption. In this paper, we present a proxy re-signature scheme based on quadratic residues, which is Bidirectional. The scheme is safety under the random oracle model. At the same time, the definition of forward-secure proxy re-signature is given, then a corresponding scheme is presented which can maintain the former signature's safety even if current cycle's Signature key has been leaked.

Japanese Journal of Applied Physics, Nov 1, 1999

The strain effect on the band structure of InAs/GaAs quantum dots has been investigated. 1 µm thi... more The strain effect on the band structure of InAs/GaAs quantum dots has been investigated. 1 µm thick InGaAs cap layer was added onto the InAs quantum dot layer to modify the strain in the quantum dots. The exciton energies of InAs quantum dots before and after the relaxation of the cap layer were determined by photoluminescence. When the epilayer was lifted off from the substrate by etching away the sacrifice layer (AlAs) by HF solution, the energy of exciton in the quantum dots decreases due to band gap narrowing resulted from the strain relaxation. This method can be used to obtain much longer emission wavelength from InAs quantum dots.