Source author record

Hassan Ali

Hassan Ali appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

Cryptography and Security Machine Learning physics.optics Computer Vision cond-mat.mes-hall cond-mat.mtrl-sci eess.IV quant-ph

Catalog footprint

What is connected

5works

8topics

4close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2022arXiv

Visual Heart Rate Estimation from RGB Facial Video using Spectral Reflectance

Estimation of the Heart rate from the facial video has a number of applications in the medical and fitness industries. Additionally, it has become useful in the field of gaming as well. Several approaches have been proposed to seamlessly obtain the Heart rate from the facial video, but these approaches have had issues in dealing with motion and illumination artifacts. In this work, we propose a reliable HR estimation framework using the spectral reflectance of the user, which makes it robust to motion and illumination disturbances. We employ deep learning-based frameworks such as Faster RCNNs to perform face detection as opposed to the Viola Jones algorithm employed by previous approaches. We evaluate our method on the MAHNOB HCI dataset and found that the proposed method is able to outperform previous approaches.Estimation of the Heart rate from facial video has a number of applications in the medical and the fitness industries. Additionally, it has become useful in the field of gaming as well. Several approaches have been proposed to seamlessly obtain the Heart rate from the facial video, but these approaches have had issues in dealing with motion and illumination artifacts. In this work, we propose a reliable HR estimation framework using the spectral reflectance of the user, which makes it robust to motion and illumination disturbances. We employ deep learning-based frameworks such as Faster RCNNs to perform face detection as opposed to the Viola-Jones algorithm employed by previous approaches. We evaluate our method on the MAHNOB HCI dataset and found that the proposed method is able to outperform previous approaches.

preprint2020arXiv

QuSecNets: Quantization-based Defense Mechanism for Securing Deep Neural Network against Adversarial Attacks

Adversarial examples have emerged as a significant threat to machine learning algorithms, especially to the convolutional neural networks (CNNs). In this paper, we propose two quantization-based defense mechanisms, Constant Quantization (CQ) and Trainable Quantization (TQ), to increase the robustness of CNNs against adversarial examples. CQ quantizes input pixel intensities based on a "fixed" number of quantization levels, while in TQ, the quantization levels are "iteratively learned during the training phase", thereby providing a stronger defense mechanism. We apply the proposed techniques on undefended CNNs against different state-of-the-art adversarial attacks from the open-source \textit{Cleverhans} library. The experimental results demonstrate 50%-96% and 10%-50% increase in the classification accuracy of the perturbed images generated from the MNIST and the CIFAR-10 datasets, respectively, on commonly used CNN (Conv2D(64, 8x8) - Conv2D(128, 6x6) - Conv2D(128, 5x5) - Dense(10) - Softmax()) available in \textit{Cleverhans} library.

preprint2020arXiv

SSCNets: Robustifying DNNs using Secure Selective Convolutional Filters

In this paper, we introduce a novel technique based on the Secure Selective Convolutional (SSC) techniques in the training loop that increases the robustness of a given DNN by allowing it to learn the data distribution based on the important edges in the input image. We validate our technique on Convolutional DNNs against the state-of-the-art attacks from the open-source Cleverhans library using the MNIST, the CIFAR-10, and the CIFAR-100 datasets. Our experimental results show that the attack success rate, as well as the imperceptibility of the adversarial images, can be significantly reduced by adding effective pre-processing functions, i.e., Sobel filtering.

preprint2016arXiv

Recombination processes in CuInS$_{2}$/ZnS Nanocrystals during steady-state photoluminescence

We report on temperature- and excitation-power-dependent photoluminescence (PL) study of CuInS$_{2}$/ZnS nanocrystals dispersed on a SiO$_{2}$/Si substrate with a confocal micro-PL system. With increasing the excitation power at 22 K and room temperature, the PL spectra are blue-shifted because of the state filling. At low temperature, a small peak is observed at the low energy side of the spectrum, which could be due to the F$\ddot{o}$rster resonance energy transfer between different nanocrystals. The integrated PL intensity increases sublinearly as a function of excitation power with a power factor of around 2/3, which demonstrates the Auger recombination dominated process in the nanocrystals, especially under the high excitation power.

preprint2015arXiv

Observation of coupling between zero- and two-dimensional semiconductor systems based on anomalous diamagnetic effects

We report the direct observation of coupling between a single self-assembled InAs quantum dot and a wetting layer, based on strong diamagnetic shifts of many-body exciton states using magneto-photoluminescence spectroscopy. An extremely large positive diamagnetic coefficient is observed when an electron in the wetting layer combines with a hole in the quantum dot; the coefficient is nearly one order of magnitude larger than that of the exciton states confined in the quantum dots. Recombination of electrons with holes in a quantum dot of the coupled system leads to an unusual negative diamagnetic effect, which is five times stronger than that in a pure quantum dot system. This effect can be attributed to the expansion of the wavefunction of remaining electrons in the wetting layer or the spread of electrons in the excited states of the quantum dot to the wetting layer after recombination. In this case, the wavefunction extent of the final states in the quantum dot plane is much larger than that of the initial states because of the absence of holes in the quantum dot to attract electrons. The properties of emitted photons that depend on the large electron wavefunction extents in the wetting layer indicate that the coupling occurs between systems of different dimensionality, which is also verified from the results obtained by applying a magnetic field in different configurations. This study paves a new way to observe hybrid states with zero- and two-dimensional structures, which could be useful for investigating the Kondo physics and implementing spin-based solid-state quantum information processing.