Dummy Output Research Articles

Protecting Intellectual Property With Reliable Availability of Learning Models in AI-Based Cybersecurity Services

Artificial intelligence (AI)-based cybersecurity services offer significant promise in many scenarios, including malware detection, content supervision, and so on. Meanwhile, many commercial and government applications have raised the need for intellectual property protection of using deep neural network (DNN). Existing studies (e.g., watermarking techniques) on intellectual property protection only aim at inserting secret information into DNNs, allowing producers to detect whether the given DNN infringes on their own copyrights. However, since the availability protection of learning models is rarely considered, the piracy model can still work with high accuracy. In this paper, a novel model locking (M-LOCK) scheme for the DNN is proposed to enhance its availability protection, where the DNN produces poor accuracy if a specific token is absent, while it maps only the tokenized inputs into correct predictions. The proposed scheme performs the verification process during the DNN inference operation, actively protecting models' intellectual property copyright at each query. Specifically, to train the token-sensitive decision-making boundaries of DNNs, a data poisoning-based model manipulation (DPMM) method is also proposed, which minimizes the correlation between the dummy outputs and correct predictions. Extensive experiments demonstrate the proposed scheme could achieve high reliability and effectiveness across various benchmark datasets as well as typical model protection methods.

Noise characterization of IUCAA digital sampling array controller

IUCAA Digital Sampling Array Controller (IDSAC) is a flexible and generic yet powerful CCD controller which can handle a wide range of scientific detectors. Based on an easily scalable modular backplane architecture consisting of Single Board Controllers (SBC), IDSAC can control large detector arrays and mosaics. Each of the SBCs offers the full functionality required to control a CCD independently. The SBCs can be cold swapped without the need to reconfigure them. Each SBC can handle data from up to four video channels with or without dummy outputs at speeds up to 500 kilo Pixels Per Second (kPPS) Per Channel with a resolution of 16 bits. Communication with Linux based host computer is through a USB3.0 interface, with the option of using copper or optical fibers. A Field Programmable Gate Array (FPGA) is used as the master controller in each SBC which allows great flexibility in optimizing performance by adjusting gain, timing signals, bias levels, etc. using user-editable configuration files without altering the circuit topology. Elimination of thermal kTC noise is achieved via Digital Correlated Double Sampling (DCDS). We present the results of noise performance characterization of IDSAC through simulation, theoretical modeling, and actual measurements. The contribution of different types of noise sources is modeled using a tool to predict noise of a generic DCDS signal chain analytically. The analytical model predicts the net input referenced noise of the signal chain to be 5 electrons for 200k pixels per second per channel readout rate with 3 samples per pixel. Using a cryogenic test set up in the lab, the noise is measured to be 5.4 e (24.3 \muV), for the same readout configuration.

Degradation of a COTS linear CCD induced by proton irradiation

A Commercial Off-The-Shelf (COTS) linear CCD irradiated by protons is examined experimentally in this study. A dummy output voltage, the charge transfer efficiency (CTE) and a dark signal were degraded by 2-, 5-, and 10-MeV proton irradiation for comparison. The CCDs irradiated by 2-MeV protons were damaged most seriously, and the CCDs irradiated by 5-MeV protons were damaged more than those irradiated by 10-MeV protons. Biased CCDs were damaged more seriously than unbiased CCDs with the same proton irradiation conditions. The vacancy density and distribution induced by proton displacement damage and the ionizing energy loss induced by proton ionization damage were calculated using TRIM software. The correlation between CTE degradation and nonionizing energy loss is analyzed.

Different Dose Rate Radiation Effects on Linear CCDs

Charge coupled devices (CCD) have been tested at widely differing dose rates to examine the radiation tolerance dependence on the dose rates. The test results show a maximum tolerance of CCDs at 10.2 rad(Si)/sec, a slight reduction in tolerance at 34.8 rad(Si)/sec and a quite precipitous roll off when moving down to 1 rad(Si)/sec and 0.1 rad(Si)/sec. The degradation of the dummy output voltages and the dark signal voltages are compared at the dose rates of 0.1,1.0,10.2 and 34.8 rad(Si)/sec, respectively. It shows that the degradation levels depend on the dose rates. The CCDs are divided into two groups during 60Co γ irradiation. Either the output amplifiers or the photo sensing and the shift register areas are shielded with Pb during the irradiation tests.

Linealizacion por realimentacion constructiva de sistemas mecanicos con grado de subactuacion 1 inestables con friccion

En los últimos años se han desarrollado diversos métodos para controlar sistemas mecánicos subactuados. De hecho, dichos métodos no lineales han conseguido resolver problemas muy interesantes desde el punto de vista de control. Sin embargo, para obtener una solución con estos métodos, normalmente es necesario resolver un sistema de ecuaciones diferenciales en derivadas parciales, lo cual no siempre es posible. En este artículo, se presenta una metodología constructiva para diseñar un controlador para una clase de sistemas mecánicos inestables en bucle abierto y con grado de subactuación uno. Para el diseño se emplean principalmente las técnicas de linealización por realimentación y Lyapunov. La metodología consiste en proponer una salida ficticia que se pueda rediseñar de forma constructiva para resolver el problema planteado, dando lugar a una ley de control explícita, que permite tener en cuenta la fricción incluso en las coordenadas no actuadas.

Design of diffractive optical elements with optimization of the signal-to-noise ratio and without a dummy area

We describe a new approach to suppress undesired diffraction orders in the signal area of a Fourier plane diffractive optical element (DOE). We implement this new approach for the DOE design by a two-stage iterative Fourier transform algorithm that incorporates an adaptive optimization of the signal-to-noise ratio and does not require the introduction of a dummy output area outside the field of view. A comparison among this approach and three other approaches are presented on the basis of numerical results from several sample diffraction patterns.