2 August 2018 in Scientific Reports
High refractive index makes silicon the optimal platform for dielectric metasurfaces capable of versatile control of light. Among various silicon modifications, its monocrystalline form has the weakest visible light absorption but requires a careful choice of the fabrication technique to avoid damage, contamination or amorphization. Presently prevailing chemical etching can shape thin silicon layers into two-dimensional patterns consisting of strips and posts with vertical walls and equal height. Here, the possibility to create silicon nanostructure of truly tree-dimensional shape by means of the focused ion beam lithography is explored, and a 300 nm thin film of monocrystalline epitaxial silicon on sapphire is patterned with a chiral nanoscale relief. It is demonstrated that exposing silicon to the ion beam causes a substantial drop of the visible transparency, which, however, is completely restored by annealing with oxidation of the damaged surface layer. As a result, the fabricated chiral metasurface combines high (50–80%) transmittance with the circular dichroism of up to 0.5 and the optical activity of up to 20° in the visible range. Being also remarkably durable, it possesses crystal-grade hardness, heat resistance up to 1000 °C and the inertness of glass.
3 October 2017 in Journal of Microscopy
10 March 2025 in IEEE Transactions on Information Forensics and Security
While Deep Neural Networks demonstrate remarkable performance in practical tasks, they are vulnerable to membership inference attacks aimed at identifying whether a certain object belongs to the training dataset. To conduct a membership inference attack on a target model, an adversary has to train a set of shadow models and conduct a statistical test to determine the membership status of the particular input object. Usually, shadow models are trained without taking into account the target model; we argue that utilizing the predictions of the target model can guide the training process of the shadow model. To improve the efficiency of shadow model-based membership inference attacks, we propose GLiRA, a knowledge distillation-guided approach to membership inference attacks. We observe that the knowledge distillation significantly improves the efficiency of a likelihood ratio membership inference attack when the architecture of the target model is both known and unknown to an attacker. We evaluate the proposed method across multiple image classification datasets and model architectures and demonstrate that knowledge distillation-guided likelihood ratio attack outperforms the current state-of-the-art membership inference attacks in the majority of experimental settings.
8 November 2021 in IEEE Control Systems Letters
Clinical infrared imagers, such as those used to visualize subcutaneous vasculature, rely on the image-projection feedback loops. Specifically, these instruments embed model-based feedback algorithms to process the acquired ’invisible’ infrared data streams and then to project a visible copy of them back to the imaged region (e.g., the skin of the patient’s forearm). Being inherently noisy, the infrared frames and their projections are prone to misalignment, demanding frequent instrument tuning and recalibration. To address this challenge, we propose to reconsider the feedback loop entailed in such imagers from the standpoint of multi-agent deep learning. Both proposed agents – Denoiser and Aligner – are adaptive in that they continuously optimize the corresponding target value functions. Namely, the Denoiser learns the proper frequency decomposition of the acquired infrared data until the target segmentation metric is maximized; whereas, the Aligner learns the intensity fluctuations within the segmentation mask tuned by the Denoiser until its maximal overlap with the source infrared image. The idea is validated synthetically on a benchmark dataset and experimentally on a bench-top vein scanner in the lab, with the duet of agents proving efficient in handling the distortions and the misalignment.
26 January 2025 in Pattern Recognition Letters
9 August 2023 in IEEE Access
State-of-the-art object detection models are frequently trained offline using available datasets, such as ImageNet: large and overly diverse data that are unbalanced and hard to cluster semantically. This kind of training drops the object detection performance should the change in illumination, in the environmental conditions (e.g., rain or dust), or in the lens positioning (out-of-focus blur) occur. We propose a simple way to intelligently control the camera and the lens focusing settings in such scenarios using DASHA, a Decentralized Autofocusing System with Hierarchical Agents. Our agents learn to focus on scenes in challenging environments, significantly enhancing the pattern recognition capacity beyond the popular detection models (YOLO, Faster R-CNN, and Retina are considered). At the same time, the decentralized training allows preserving the equipment from overheating. The algorithm relies on the latent representation of the camera’s stream and, thus, it is the first method to allow a completely no-reference imaging, where the system trains itself to auto-focus itself.