arXiv subjects

Sarwar Khan

Publications and source records attributed to Sarwar Khan.

8 recordsLinked to original sources

arXiv2026

First detection of C2H+ in the interstellar medium

Despite the detection of nearly 350 molecules in the interstellar medium, almost half of which are carbon chains, the pathways that build molecular complexity remain poorly understood. Observed abundances of carbon-chain and aromatic species are difficult to reconcile with existing top-down or bottom-up formation scenarios, due in part to limited observational constraints and incomplete theoretical understanding. In particular, small intermediary ions, key drivers of ion-molecule reactions capable of seeding larger hydrocarbons and aromatic rings, could provide critical support for the bottom-up formation scenario. Constraining the abundance and chemistry of these ions is therefore essential to test whether bottom-up growth can operate efficiently under interstellar conditions. Here, we report the first detection of the small hydrocarbon cation ethynylium, C2H+, toward the Orion Bar, based on observations with the APEX 12m sub-mm telescope of its lowest-lying J=3-2 rotational transition near 211GHz, which exhibits a unique spectroscopic fingerprint through resolved Lambda-doubling and hyperfine splitting components, as recently measured in the laboratory. Meudon PDR models successfully reproduce these values, placing C2H+ formation at the outer edges of PDR fronts. Our results link C2H+ production to CH+ and CH3+ within a network of ion-molecule reactions driven by vibrationally excited H2, a scenario now further supported by recent detections of these species in PDRs like the Orion Bar with JWST observations. The importance of C2H+ lies in its role as a key intermediate: it produces C2H2+ and subsequently C2H3+, effectively channelling small C2 building blocks toward larger hydrocarbons and facilitating bottom-up growth at the PDR surface. Targeted searches for C2H+ in other regions promise to provide a potentially decisive probe of ion-driven bottom-up chemistry in the ISM.

astro-ph.GA↗

arXiv2024

A multi-wavelength study of Galactic H II regions with extended emission

H II regions are the signposts of massive ($M\geq\,8\,M_\odot$) star-forming sites in our Galaxy. It has been observed that the ionizing photon rate inferred from the radio continuum emission of H II regions is significantly lower ($\sim$ 90%) than that inferred from far-infrared fluxes measured by IRAS. This discrepancy in the ionizing photon rates may arise due to there being significant amounts of dust within the H II regions or the presence of extended emission that is undetected by high-resolution radio interferometric observations. Here, we study a sample of eight compact and ultracompact H II regions with extended emission to explore its role in resolving the discrepancy. We have used observations at the uGMRT (1.25-1.45 GHz) and data from the GLOSTAR survey (4-8 GHz) to estimate the ionizing photon rate from the radio continuum emission. We have also estimated the ionizing photon rate from the infrared luminosity by fitting a spectral energy distribution function to the infrared data from the GLIMPSE, MIPSGAL, and Hi-GAL surveys. The excellent sensitivity of the radio observations to extended emission allows us to investigate the actual fraction of ionizing photons that are absorbed by dust in compact and ultracompact H II regions. Barring one source, we find a direct association between the radio continuum emission from the compact and diffuse components of the H II region. Our study shows that the ionizing photon rates estimated using the radio and infrared data are within reasonable agreement (5-28%) if we include the extended emission. We also find multiple candidate ionizing stars in all our sources, and the ionizing photon rates from the radio observations and candidate stars are in reasonable agreement.

astro-ph.GA↗

arXiv2024

Adversarially Robust Deepfake Detection via Adversarial Feature Similarity Learning

Deepfake technology has raised concerns about the authenticity of digital content, necessitating the development of effective detection methods. However, the widespread availability of deepfakes has given rise to a new challenge in the form of adversarial attacks. Adversaries can manipulate deepfake videos with small, imperceptible perturbations that can deceive the detection models into producing incorrect outputs. To tackle this critical issue, we introduce Adversarial Feature Similarity Learning (AFSL), which integrates three fundamental deep feature learning paradigms. By optimizing the similarity between samples and weight vectors, our approach aims to distinguish between real and fake instances. Additionally, we aim to maximize the similarity between both adversarially perturbed examples and unperturbed examples, regardless of their real or fake nature. Moreover, we introduce a regularization technique that maximizes the dissimilarity between real and fake samples, ensuring a clear separation between these two categories. With extensive experiments on popular deepfake datasets, including FaceForensics++, FaceShifter, and DeeperForensics, the proposed method outperforms other standard adversarial training-based defense methods significantly. This further demonstrates the effectiveness of our approach to protecting deepfake detectors from adversarial attacks.

cs.CV↗

arXiv2023

CapST: Leveraging Capsule Networks and Temporal Attention for Accurate Model Attribution in Deep-fake Videos

Deep-fake videos, generated through AI face-swapping techniques, have gained significant attention due to their potential for impactful impersonation attacks. While most research focuses on real vs. fake detection, attributing a deep-fake to its specific generation model or encoder is vital for forensic analysis, enabling source tracing and tailored countermeasures. This enhances detection by leveraging model-specific artifacts and supports proactive defenses. We investigate the model attribution problem for deep-fake videos using two datasets: Deepfakes from Different Models (DFDM) and GANGen-Detection, both comprising deep-fake videos and GAN-generated images. We use only fake images from GANGen-Detection to align with DFDM's focus on attribution rather than binary classification. We formulate the task as a multiclass classification problem and introduce a novel Capsule-Spatial-Temporal (CapST) model that integrates a truncated VGG19 network for feature extraction, capsule networks for hierarchical encoding, and a spatio-temporal attention mechanism. Video-level fusion captures temporal dependencies across frames. Experiments on DFDM and GANGen-Detection show CapST outperforms baseline models in attribution accuracy while reducing computational cost.

cs.CV↗

arXiv2022

Macromolecule Classification Based on the Amino-acid Sequence

Deep learning is playing a vital role in every field which involves data. It has emerged as a strong and efficient framework that can be applied to a broad spectrum of complex learning problems which were difficult to solve using traditional machine learning techniques in the past. In this study we focused on classification of protein sequences with deep learning techniques. The study of amino acid sequence is vital in life sciences. We used different word embedding techniques from Natural Language processing to represent the amino acid sequence as vectors. Our main goal was to classify sequences to four group of classes, that are DNA, RNA, Protein and hybrid. After several tests we have achieved almost 99% of train and test accuracy. We have experimented on CNN, LSTM, Bidirectional LSTM, and GRU.

q-bio.BM↗

arXiv2022

Classification of Macromolecule Type Based on Sequences of Amino Acids Using Deep Learning

The classification of amino acids and their sequence analysis plays a vital role in life sciences and is a challenging task. This article uses and compares state-of-the-art deep learning models like convolution neural networks (CNN), long short-term memory (LSTM), and gated recurrent units (GRU) to solve macromolecule classification problems using amino acids. These models have efficient frameworks for solving a broad spectrum of complex learning problems compared to traditional machine learning techniques. We use word embedding to represent the amino acid sequences as vectors. The CNN extracts features from amino acid sequences, which are treated as vectors, then fed to the models mentioned above to train a robust classifier. Our results show that word2vec as embedding combined with VGG-16 performs better than LSTM and GRU. The proposed approach gets an error rate of 1.5%.

q-bio.BM↗

arXiv2022

A multiwavelength study of the W33 Main ultracompact HII region

The dynamics of ionized gas around the W33 Main ultracompact HII region is studied using observations of hydrogen radio recombination lines and a detailed multiwavelength characterization of the massive star-forming region W33 Main is performed. We used the Giant Meterwave Radio Telescope (GMRT) to observe the H167$α$ recombination line at 1.4 GHz at an angular resolution of 10 arcsec, and Karl. G. Jansky Very Large Array (VLA) data acquired in the GLOSTAR survey to study the dynamics of ionized gas. We also observed the radio continuum at 1.4 GHz and 610 MHz with the GMRT and used GLOSTAR 4$-$8 GHz continuum data to characterize the nature of the radio emission. In addition, archival data from submillimeter to near-infrared wavelengths were used to study the dust emission and identify YSOs in the W33 Main star-forming region. The radio recombination lines were detected at good signal to noise in the GLOSTAR data, while the H167$α$ radio recombination line was marginally detected with the GMRT. The spectral index of radio emission in the region determined from GMRT and GLOSTAR shows the emission to be thermal in the entire region. Along with W33 Main, an arc-shaped diffuse continuum source, G12.81$-$0.22, was detected with the GMRT data. The GLOSTAR recombination line data reveal a velocity gradient across W33 Main and G12.81$-$0.22. The electron temperature is found to be 6343 K and 4843 K in W33 Main and G12.81$-$0.22, respectively. The physical properties of the W33 Main molecular clump were derived by modeling the dust emission using data from the ATLASGAL and Hi-GAL surveys and they are consistent with the region being a relatively evolved site of massive star formation. The gas dynamics and physical properties of G12.81$-$0.22 are consistent with the HII region being in an evolved phase and its expansion on account of the pressure difference is slowing down.

astro-ph.GA↗

arXiv2020

Detection of Diabetic Anomalies in Retinal Images using Morphological Cascading Decision Tree

This research aims to develop an efficient system for screening of diabetic retinopathy. Diabetic retinopathy is the major cause of blindness. Severity of diabetic retinopathy is recognized by some features, such as blood vessel area, exudates, haemorrhages and microaneurysms. To grade the disease the screening system must efficiently detect these features. In this paper we are proposing a simple and fast method for detection of diabetic retinopathy. We do pre-processing of grey-scale image and find all labelled connected components (blobs) in an image regardless of whether it is haemorrhages, exudates, vessels, optic disc or anything else. Then we apply some constraints such as compactness, area of blob, intensity and contrast for screening of candidate connectedcomponent responsible for diabetic retinopathy. We obtain our final results by doing some post processing. The results are compared with ground truths. Performance is measured by finding the recall (sensitivity). We took 10 images of dimension 500 * 752. The mean recall is 90.03%.

eess.IV↗