Many pretrained deep learning models have been released to help engineers and researchers develop deep learning-based systems or conduct research with minimall effort. Previous work has shown that at secret message can be embedded in neural network parameters without compromising the accuracy of the model. Malicious developers can, therefore, hide malware or other baneful information in pretrained models, causing harm to society. Hence, reliable detection of these vicious pretrained models is urgently needed. We analyze existing approaches for hiding messages and find that they will ineluctably cause biases in the parameter statistics. Therefore, we propose steganalysis methods for steganography on neural network parameters that extract statistics from benign and malicious models and build classifiers based on the extracted statistics. To the best of our knowledge, this is the first study on neural network steganalysis. The experimental results reveal that our proposed algorithm can effectively detect a model with an embedded message. Notably, our detection methods are still valid in cases where the payload of the stego model is low.

Ultraviolet photodetection plays an important role in optical communication and chemical- and bio- related sensing applications. Gallium nitride (GaN) nanowires-based photoelectrochemical-type photodetectors, which operate particularly in acqueous conditions, have been attracted extensive interest because of their low cost, fast photoresponse, and excellent responsivity. However, GaN nanowires, which have a large surface-to-volume ratio, suffer suffered from instability in photoelectrochemical environments because of photocorrosion. In this study, the structural and photoelectrochemical properties of GaN nanowires with improved photoresponse and chemical stability obtained by coating the nanowire surface with an ultrathin TiO₂ protective layer were investigated. The photocurrent density of TiO₂-coated GaN nanowires changed minimally over a relatively long operation time of 2000 s under 365-nm illumination. Meanwhile, the attenuation coefficient of the photocurrent density could be reduced to 49%, whereas it is as high as 85% in uncoated GaN nanowires. Furthermore, the photoelectrochemical behavior of the nanowires was investigated through electrochemical impedance spectroscopy measurements. The results shed light on the construction of long-term-stable GaN-nanowire-based photoelectrochemical-type photodetectors.

The high granularity timing detector (HGTD) is a crucial component of the ATLAS phase II upgrade to cope with the extremely high pile-up (the average number of interactions per bunch crossing can be as high as 200). With the precise timing information (σ_t～30 ps) of the tracks, the track-to-vertex association can be performed in the “4-D” space. The Low Gain Avalanche Detector (LGAD) technology is chosen for the sensors, which can provide the required timing resolution and good signal-to-noise ratio. Hamamatsu Photonics K.K. (HPK) has produced the LGAD with thicknesses of 35 μm and 50 μm. The University of Science and Technology of China(USTC) has also developed and produced 50 μm LGADs prototypes with the Institute of Microelectronics (IME) of Chinese Academy of Sciences. To evaluate the irradiation hardness, the sensors are irradiated with the neutron at the JSI reactor facility and tested at USTC. The irradiation effects on both the gain layer and the bulk are characterized by I-V and C-V measurements at room temperature (20 ℃) or −30 ℃. The breakdown voltages and depletion voltages are extracted and presented as a function of the fluences. The final fitting of the acceptor removal model yielded the c-factor of 3.06×10⁻¹⁶ cm⁻², 3.89×10⁻¹⁶ cm⁻² and 4.12×10⁻¹⁶ cm⁻² for the HPK-1.2, HPK-3.2 and USTC-1.1-W8, respectively, showing that the HPK-1.2 sensors have the most irradiation resistant gain layer. A novel analysis method is used to further exploit the data to get the relationship between the c-factor and initial doping density.

The spray cold plate has a compact structure and high-efficiency heat exchange, which can meet the requirements of high heat flux dissipation of multiple heat sources, and is a reliable means to solve the heat dissipation of the next generation of chips. This paper proposes to use surfactants to enhance the heat transfer of the spray cold plate, and conduct a systematic experimental study on the heat transfer performance of the spray cold plate under different types and concentrations of additives. It was found that among the three surfactants, sodium dodecyl sulfate (SDS) can improve the heat transfer performance of the spray cold plate, and at the optimal concentration of 200ppm, the heat transfer coefficient of the spray cold plate was increased significantly by 19.8%. Both the n-octanol-distilled water and Tween 20-distilled water can reduce the heat transfer performance of the cold plate using multi nozzles. In addition, based on the experimental data, the dimensionless heat transfers correlations for the spray cold plate using additives were conducted, and the maximum errors of dimensionless correlations for using additives were 2.1%, 2.8%, and 5.4% respectively. This discovery provides a theoretical analysis and basis for the improvement of spray cold plates.

Liquid Li metal is a promising nuclear reactor coolant; however, relevant research regarding its heat transfer characteristics remains insufficient. In this study, a steady-state two-dimensional mathematical model is established to describe the heat transfer process of liquid Li in a straight pipe. A numerical analysis is conducted to investigate the effects of inlet velocity, inlet temperature, and wall heat flux on heat transfer in liquid Li. The results indicate the advantage of using liquid Li for improving heat transfer at high inlet temperatures (> 1000 K) compared with using liquid sodium and lead–bismuth eutectic. Considering the mechanism of the outlet radial heat flow model, the ratio of turbulent to molecular diffusion coefficients presents a parabolic distribution along the radius of the pipe. Increasing the inlet velocity, decreasing the inlet temperature, and decreasing the wall heat flux can effectively weaken the dominant role of molecular heat transfer owing to the low Prandtl number of liquid Li. The heat transfer of liquid Li is investigated comprehensively in this study, and the results provide a basis for the practical application of liquid Li as a promising coolant.

The energy consumption problem of heating, ventilation, and air conditioning systems over general directed graphs is investigated. The considered problem is firstly reformulated as a Nash equilibrium seeking problem, and a distributed consensus-based algorithm is then proposed to solve it. To address the challenge arising from general directed graphs, a distributed estimation algorithm is embedded such that the explicit dependence on the left eigenvector associated with the eigenvalue zero of the Laplacian matrix can be avoided. Then, the exponential convergence of the proposed distributed Nash equilibrium seeking algorithm is established under a standing assumption. A numerical example is finally provided to verify the effectiveness of the proposed algorithm.

Clarifying the mechanism of fungi growth is of great significance for maintaining the quality during grain storage. Among the factors that affect the growth of fungi spores, the most important factors are temperature, moisture content and storage time. Therefore, through this study, a multivariate linear regression model among several important factors, such as the spore number and ambient temperature, rice moisture content and storage days, were developed based on the experimental data. In order to build a more accurate model, we introduce a random forest algorithm into the fungal spore prediction during grain storage. The established regression models can be used to predict the spore number under different ambient temperature, rice moisture content and storage days during the storage process. For the random forest model, it could control the predicted value to be of the same order of magnitude as the actual value for 99% of the original data, which have a high accuracy to predict the spore number during the storage process. Furthermore, we plot the prediction surface graph to help practitioners to control the storage environment within the conditions in the low risk region.

Sound speed is an important acoustic parameter for tissue characterization. Herein we developed an ultrasound computed tomography (USCT) system for ex vivo sound speed imaging and evaluation of small animal organs. The proposed USCT system employs a 256-element ring array transducer and allows simultaneous signal transmission and reception for all channels. The method does not require complicated sample preparation procedures and can yield accurate measurement results. Experimental results show that sound speeds of excised rat brain, heart, liver, spleen, and kidney measured by the method are close to published data. This work demonstrates a new method for sound speed imaging and holds potential for in vivo applications.