Department of Computer Science

Ashoka University is a private university located in Sonipat, Haryana, India, focusing on liberal arts and sciences. It was established in 2014.&nbsp;It is guided by an academic council of academics and scholars from around the world. The council sets the university&#39;s academic standard, advising on matters of curriculum, faculty hiring and research.&nbsp;

Ashoka University offers a full-time 3 (+ 1) year undergraduate programme. The University comprises of&nbsp;2500 plus students, with a mix of both national and international students.&nbsp; &nbsp;

&nbsp;

&nbsp;

Ashoka University

CLIPStyler demonstrated image style transfer with realistic textures using only the style text description (instead of requiring a reference style image). However, the ground semantics of objects in style transfer output is lost due to style spillover on salient and background objects (content mismatch) or over-stylization. To solve this, we propose Semantic CLIPStyler (Sem-CS) that performs semantic style transfer. Sem-CS first segments the content image into salient and non-salient objects and then transfers artistic style based on a given style text description. The semantic style transfer is achieved using global foreground loss (for salient objects) and global background loss (for non-salient objects). Our empirical results, including DISTS, NIMA and user study scores, show that our proposed framework yields superior qualitative and quantitative performance.

<a href="https://arxiv.org/abs/2303.06334">SEM-CS: Semantic CLIPStyler for Text-Based Image Style Transfer</a>

Performing computation in the presence of faulty and malicious nodes is a central problem in distributed computing. Over 35 years ago, Dwork, Peleg, Pippenger, and Upfal [STOC 1986, SICOMP 1988] studied the fundamental Byzantine agreement problem in sparse, bounded degree networks and presented the first protocol that achieved almost-everywhere agreement among good nodes. However, this protocol and several subsequent protocols including that of King, Saia, Sanwalani, and Vee [FOCS 2006] had the drawback that they were not fully-distributed - in those protocols, nodes are required to have initial knowledge of the entire network topology. This drawback makes such protocols not applicable to real-world communication networks such as peer-to-peer (P2P) networks, which are typically sparse and bounded degree and where nodes initially have only local knowledge of themselves and of their neighbors.

World University root node

Annual ACM Symposium on Parallelism in Algorithms and Architectures

A Fully-Distributed Scalable Peer-to-Peer Protocol for Byzantine-Resilient Distributed Hash Tables

While detection of malignancies on mammography has received a boost with the use of Convolutional Neural Networks (CNN), detection of cancers of very small size remains challenging. This is however clinically significant as the purpose of mammography is early detection of cancer, making it imperative to pick them up when they are still very small. Mammography has the highest spatial resolution (image sizes as high as 3328&nbsp;&times;&nbsp;4096 pixels) out of all imaging modalities, a requirement that stems from the need to detect fine features of the smallest cancers on screening. However due to computational constraints, most state of the art CNNs work on reduced resolution images. Those that work on higher resolutions, compromise on global context and work at single scale. In this work, we show that resolution, scale and image-context are all important independent factors in detection of small masses. We thereby use a fully convolutional network, with the ability to take any input size. In addition, we incorporate a systematic multi-scale, multi-resolution approach, and encode image context, which we show are critical factors to detection of small masses. We show that this approach improves the detection of cancer, particularly for small masses in comparison to the baseline model. We perform a single institution multicentre study, and show the performance of the model on a diagnostic mammography dataset, a screening mammography dataset, as well as a curated dataset of small cancers &lt; 1&nbsp;cm in size. We show that our approach improves the sensitivity from 61.53 to 87.18% at 0.3 False Positives per Image (FPI) on this small cancer dataset. Model and code are available from&nbsp;https://github.com/amangupt01/Small_Cancer_Detection.

Fulltext

Scientific Reports

Ultra-high resolution, multi-scale, context-aware approach for detection of small cancers on mammography

India confines more than 17% of the world&rsquo;s population and has a diverse genetic makeup with several clinically relevant rare mutations belonging to many sub-group which are undervalued in global sequencing datasets like the 1000 Genome data (1KG) containing limited samples for Indian ethnicity. Such databases are critical for the pharmaceutical and drug development industry where diversity plays a crucial role in identifying genetic disposition towards adverse drug reactions. A qualitative and comparative sequence and structural study utilizing variant information present in the recently published, largest curated Indian genome database (IndiGen) and the 1000 Genome data was performed for variants belonging to the kinase coding genes, the second most targeted group of drug targets. The sequence-level analysis identified similarities and differences among different populations based on the nsSNVs and amino acid exchange frequencies whereas a comparative structural analysis of IndiGen variants was performed with pathogenic variants reported in UniProtKB Humsavar data. The influence of these variations on structural features of the protein, such as structural stability, solvent accessibility, hydrophobicity, and the hydrogen-bond network was investigated. In-silico screening of the known drugs to these Indian variation-containing proteins reveals critical differences imparted in the strength of binding due to the variations present in the Indian population. In conclusion, this study constitutes a comprehensive investigation into the understanding of common variations present in the second largest population in the world and investigating its implications in the sequence, structural and pharmacogenomic landscape. The preliminary investigation reported in this paper, supporting the screening and detection of ADRs specific to the Indian population could aid in the development of techniques for pre-clinical and post-market screening of drug-related adverse events in the Indian population. Copyright &copy; 2022 Panda, Mishra, Sharma, Kutum, Bhoyar, Jain, Imran, Senthilvel, Divakar, Mishra, Garg, Banerjee, Sivasubbu, Scaria and Ray.

Frontiers in Pharmacology

Comprehensive Assessment of Indian Variations in the Druggable Kinome Landscape Highlights Distinct Insights at the Sequence, Structure and Pharmacogenomic Stratum

The global efforts to control COVID-19 are threatened by the rapid emergence of novel SARS-CoV-2 variants that may display undesirable characteristics such as immune escape, increased transmissibility or pathogenicity. Early prediction for emergence of new strains with these features is critical for pandemic preparedness. We present Strainflow, a supervised and causally predictive model using unsupervised latent space features of SARS-CoV-2 genome sequences. Strainflow was trained and validated on 0.9 million sequences for the period December, 2019 to June, 2021 and the frozen model was prospectively validated from July, 2021 to December, 2021. Strainflow captured the rise in cases 2&nbsp;months ahead of the Delta and Omicron surges in most countries including the prediction of a surge in India as early as beginning of November, 2021. Entropy analysis of Strainflow unsupervised embeddings clearly reveals the explore-exploit cycles in genomic feature-space, thus adding interpretability to the deep learning based model. We also conducted codon-level analysis of our model for interpretability and biological validity of our unsupervised features. Strainflow application is openly available as an interactive web-application for prospective genomic surveillance of COVID-19 across the globe.

Frontiers in Genetics

Genomic Surveillance of COVID-19 Variants With Language Models and Machine Learning

Deep Neural Networks (DNNs) are often criticized for being susceptible to adversarial attacks. Most successful defense strategies adopt adversarial training or random input transformations that typically require retraining or finetuning the model to achieve reasonable performance. In this work, our investigations of intermediate representations of a pre-trained DNN lead to an interesting discovery pointing to intrinsic robustness to adversarial attacks. We find that we can learn a generative classifier by statistically characterizing the neural response of an intermediate layer to clean training samples. The predictions of multiple such intermediate-layer based classifiers, when aggregated, show unexpected robustness to adversarial attacks. Specifically, we devise an ensemble of these generative classifiers that rank-aggregates their predictions via a Borda count-based consensus. Our proposed approach uses a subset of the clean training data and a pre-trained model, and yet is agnostic to network architectures or the adversarial attack generation method. We show extensive experiments to establish that our defense strategy achieves state-of-the-art performance on the ImageNet validation set.