Privacy-preserving data mining has become an important topic. People have built several multi-party-computation (MPC)-based frameworks to provide theoretically guaranteed privacy, the poor performance of real-world algorithms have always been a challenge. Using Principal Component Analysis (PCA) as an example, we show that by considering the unique performance characters of the MPC platform, we can design highly effective algorithm-level optimizations, such as replacing expensive operators and batching up. We achieve about 200$\times$ performance boost over existing privacy-preserving PCA algorithms with the same level of privacy guarantee. Also, using real-world datasets, we show that by combining multi-party data, we can achieve better training results.
Purpose: To quantify the morphology of eyeballs with posterior staphyloma (PS) with Zernike decomposition and to explore the association between Zernike coefficients with existing PS classification. Methods: Fifty-three eyes with high myopia (HM, ≤-6.00D) and 30 with PS were included. PS was classified with traditional methods based on OCT findings. Eyeballs’ morphology was obtained with 3D MRI, from which the height map of the posterior surface was extracted. Zernike decomposition was performed to derive the coefficients of the 1st-27th items, which were compared between HM and PS eyes with the Mann-Whitney-U test. Receiver operating characteristics (ROC) analysis was used to test the effectiveness of using Zernike coefficients to discriminate PS from HM. Results: Compared to HM eyeballs, PS eyeballs had significantly increased vertical and horizontal tilt, oblique astigmatism, defocus, vertical and horizontal coma, and higher order aberrations (HOA) (all Ps < 0.05). HOA was the most effective in PS classification with an area under the ROC curve (AUROC) value of 0.977. Among the 30 PS, 19 were the wide macular type with large defocus and negative spherical aberration; 4 were the narrow macular type with positive spherical aberration; 3 were inferior PS with greater vertical tilt, and 4were peripapillary PS with larger horizontal tilt. Conclusion: PS eyes have significantly increased Zernike coefficients, and HOA is the most effective parameter to differentiate PS from HM. The geometrical meaning of the Zernike components showed great accordance with PS classification.
Regulatory T cells (Tregs) display great promise in rheumatoid arthritis (RA) therapy. However, their low number and differentiation rate limit their further application in the clinics. In the present study, we first optimized a combination of IL-2, TGF-β and cyclin dependent kinase inhibitor AS2863619 (IL-2/TGF-β/AS), which could induce Tregs with high efficiency in vitro. After the induced Tregs (iTregs) were confirmed to suppress lymphocyte proliferation and pro-inflammatory T help cells (Th1 and Th17) activation, a chitosan-stabilized nanoparticle drug delivery system (NDDS) was developed according to the optimized formula of IL-2/TGF-β/AS. In vivo study, the NDDS was injected into the knees of mice with collagen-induced arthritis (CIA). As a result, the NDDS remarkably reduced the pathological score of the CIA, alleviated the inflammatory cell infiltration and synovial hyperplasia, and minimized cartilage tissue damage in the knee joint of the CIA mice. Mechanically, the NDDS administration promoted Treg differentiation and decreased Th17 production, consequently reversing the ratio of Treg/Th17, and reducing the secretion of TNF-α in the sera, which facilitated to relieve the severity and progression of arthritis. In sum, NDDS capable of efficiently inducing Tregs were constructed successfully and provided a potential platform for treating RA by restoring the equilibrium of Treg/Th17 destroyed in RA.
Animals and plants have numerous active protections for adapting to the complex and severe living environments, providing endless inspiration for extending the service life of materials and machines. Conch, a marine animal living near the coast and chronically suffering from the erosion of sand in water, has adapted to the condition through its anti-erosion conch shell. Romanesco broccoli, a plant whose inflorescence is self-similar in character, has a natural fractal bud’s form. Coupling the convex domes on the conch shell and the fractal structure of Romanesco broccoli, a novel valve core structure of a water hydraulic valve was designed in this paper to improve the particle erosion resistance and valve core’s service life. Three models were built to compare the effect among the normal structure, bionic structure, and multi-source coupling bionic structures, and were coined using 3D printing technology. A 3D printed water hydraulic valve was manufactured to simulate the working condition of a valve core under sand erosion in water flow, and capture the experimental videos of the two-phase flow. Furthermore, based on the water hydraulic platform and one-camera-six-mirror 3D imaging subsystem, the experiment system was established and used to compare the performance of the three different valve cores. As a result, the results showed that the coupling bionic structure could effectively improve the anti-erosion property of the valve core and protect the sealing face on the valve core from wear. This paper presents a novel way of combining advantages from both animal (function bionic) and plant (shape bionic) in one component design.
Because of general chaos and dominant position of barbarian in Middle Ages, theChristian church was the heir to Roman civilization in some extent. It played animportant role in the process of spreading civilization and cultivating Germanic people. Meanwhile, it also built up supreme authority in French. With the development of society and economy, its authority met with challenge from upper andlower society.
In recent years, with the continuous enhancement of contract management to enterprise, the improvement of business integration management level, as well as the rapid development of information technology and the Centralized planning of informatization construction of the group of companies, the current contract management system can no longer fully fulfill the requirements of the business from contract management. In order to enhance the intelligence ability of contract audit and management, the research of contract audit and management methods based on artificial intelligence have been proposed. In the aspect of contract audit based on artificial intelligence, the paper proposes to construct the business architecture to contract audit, clarify the contract audit business processing process, and provide the function of auxiliary audit. The process of contract management based on artificial intelligence takes risk early warning as the core index for management, designs an intelligent question and answer system based on artificial intelligence, and finally realizes the contract management based on artificial intelligence.
As intelligent systems that consider the driver in the loop (DiL), advanced driver-assistance systems (ADASs) have been widely applied in intelligent vehicles to improve driving safety and comfort. To provide more driving pleasure while maintaining driving safety, this paper researches the ADAS platform and proposes a human-road dual trust (HRDT) mechanism to achieve the adaptive distributed shared control under the lane keeping scenario. Differing from the existing trust mechanisms that only focus on driving information, road geometric information is introduced to assess the driving risk more comprehensively by using the fuzzy logic algorithm in the proposed mechanism. Moreover, a novel indicator is designed to provide the integrated measurement of the lateral deviation and the autopilot dependence. Finally, a hierarchical validation process with the driver-in-the-loop platform is applied to prove that the HRDT mechanism helps reduce the autopilot dependence significantly while keeping a minor lateral deviation to maintain driving safety.
This paper expounds that the system of subject librarians is necessary result of university library service innovation, discusses the station and duty of subject librarians, proposes the strategies of practicing the system of subject librarians in university library from aspects of clear-cut cultivating target, implementing cultivating measures, norm management mode, etc.