Abstract

Generative AI has seen remarkable growth over the past few years, with diffusion models being state-of-the-art for image generation. This study investigates the use of diffusion models in artificial data generation for electronic circuits to enhance the accuracy of subsequent machine learning models in tasks such as performance assessment, design, and testing when training data is usually known to be very limited. We utilize simulations in the HSPICE design environment with 22nm CMOS technology nodes to obtain representative real training data for our proposed diffusion model. Our results demonstrate the close resemblance of synthetic data using diffusion models to real data. We validate the quality of generated data and demonstrate that data augmentation is certainly effective in the predictive analysis of VLSI design for digital circuits.

Abstract

Water quality monitoring of reservoirs is currently a signifcant challenge in the tropical regions of the world due to limited monitoring stations and hydrological data. Remote sensing techniques have proven to be a powerful tool for continuous real-time monitoring and assessment of tropical reservoirs water quality. Although many studies have detected chlorophyll-a (Chl-a) concentrations as a proxy to represent nutrient contamination, using Sentinel 2 for eutrophic or hypereutrophic inland water bodies, mainly reservoirs, minimal eforts have been made for oligotrophic and mesotrophic reservoirs. The present study aimed to develop a modeling framework to map and estimate spatio-temporal variability of Chl-a levels and associated water spread using the Modifed Normalized Diference Water Index (MNDWI) and Maximum Chlorophyll Index (MCI). Moreover, the impact of land use/land cover type of the contributing watershed in the oligo-mesotrophic reservoir, Bhadra (tropical reservoir), for 2018 and 2019 using Sentinel 2 satellite data was analyzed. The results show that the water spread area was higher in the post-monsoon months and lower in the summer months. This was further validated by the correlation with reservoir storage, which showed a strong relationship (R2 = 0.97, 2018; R2 = 0.93, 2019). The estimated Chl-a spread was higher in the winter season, because the reservoir catchment was dominated by deciduous forest, producing a large amount of leaf litter in tropical regions, which leads to an increase in the level of Chl-a. It was found that Chl-a spread in the reservoir, specifcally at the inlet sources and near agricultural land practices (western parts of the Bhadra reservoir). Based on the fndings of this study, the MCI spectral index derived from Sentinel 2 data can be used to accurately map the spread of Chl-a in diverse water bodies, thereby ofering a robust scientifc basis for efective reservoir management.

Abstract

The proposed work endeavors to support the preservation of tribal arts in their original art form with their unique colors, shapes, and objects. Tribal arts of each region of India is a ‘message’ of their life, traditions, and history, requiring preservation in its pristine, unique form. For example, dotted patterns in Bhil, geometric figures in Warli, dashes in Gond, deities’ expressions in Pichwai paintings, etc., are some notably diverse visual patterns that depict the profound stories of their wholesome culture. In this preliminary work, a dataset is collected by mining the websites, which are direct outlets of tribal arts, thereby ensuring the authenticity of the images included. The collected dataset includes 863 paintings across five different classes of tribal arts viz Warli, Bhil, Gond, Pichwai, and Kalighat, as described in Figure 1. Furthermore, image classification baselines on the collected dataset are developed by fine-tuning the pre-trained ResNet50, InceptionV3, and EfficientNetB0 models. The classification accuracy for the aforementioned baselines is reported in Table 1. Among the baselines, the best testing accuracy of 91.57 % is observed for EfficientNetB0; the confusion matrix for which is shown in Figure 2. The associated F1-scores for this baseline model across all five classes, namely Warli, Bhil, Gond, Pichwai, and Kalighat, were found to be 0.958, 0.882, 0.875, 0.909, and 0.958, respectively which suggests that the dataset is balanced in nature. Furthermore, class activation maps are generated for each image of the collected dataset across all baselines to ascertain the alignment of results generated with the ground truth, which offers a means to comprehend and compare the regions of focus in decision-making. Figure 3 provides an example of a class activation map revealing the salient focus areas, such as human figures represented by triangles and circles in a Warli painting. This analysis can also be used by artists for designing the artwork. The extension of the classification work is to detect counterfeit tribal paintings and quantify the extent of deviation from the original art form, ultimately contributing towards tribal arts preservation and safeguarding their invaluable cultural heritage.

Abstract

No Results Found

Abstract

AI is about learning algorithms and huge amounts of data and are drivers of economic growth -- what does this mean for the field of development studies? Can we re-orient to twin AI studies and development theory and practice to generate how development challenges are identified and researched? To do this a good grasp is needed of AI internal mechanisms and outcomes in addressing development issues -- this argument will be developed through a case study of the ADAS [Advanced Driver Assistance System] deployment in India. Over and above discussing the ADAS we bring an anthropological lens to understand the social context that surrounds the system. Focusing on bus drivers, we offer findings from a qualitative and ethnographic study of drivers in a collaborative effort to achieve road safety by deploying AI-driven technology and empowering stakeholders in the transport industry in India especially, bus drivers as critical actors in the city transport network.

Abstract

We study communication models for channels with erasures in which the erasure pattern can be controlled by an adversary with partial knowledge of the transmitted codeword. In particular, we design block codes for channels with binary inputs with an adversary who can erase a fraction p of the transmitted bits. We consider causal adversaries, who must choose to erase an input bit using knowledge of that bit and previously transmitted bits, and myopic adversaries, who can choose an erasure pattern based on observing the transmitted codeword through a binary erasure channel with random erasures. For both settings we design efficient (polynomial time) encoding and decoding algorithms that use randomization at the encoder only. Our constructions achieve capacity for the causal and "sufficiently myopic" models. For the "insufficiently myopic" adversary, the capacity is unknown, but existing converses show the capacity is zero for a range of parameters. For all parameters outside of that range, our construction achieves positive rates.

Abstract

This paper addresses the challenges faced by Indian languages in leveraging deep learning for natural language processing (NLP) due to limited resources, annotated datasets, and Transformer-based architectures. We specifically focus on Telugu and aim to construct a Telugu morph analyzer dataset comprising 10,000 sentences. Furthermore, we assess the performance of established multilingual Transformer models (m-Bert, XLMR, IndicBERT) and mono-lingual Transformer model BERT-Te (trained from scratch on an extensive Telugu corpus comprising 80,15,588 sentences). Our findings demonstrate the efficacy of Transformer-based representations pre-trained on Telugu data improved the performance of the Telugu morph analyzer, surpassing existing multi-lingual approaches. This highlights the necessity of developing dedicated corpora, annotated datasets, and machine learning models in a mono-lingual setting. Using our dataset, we present benchmark results for the Telugu morph analyzer achieved through simple fine-tuning on BERT-Te. The morph analyzer dataset 1 and codes are open-sourced and available here.

Abstract

Brain stroke has become a significant burden on global health and thus we need remedies and prevention strategies to overcome this challenge. For this, the immediate identification of stroke and risk stratification is the primary task for clinicians. To aid expert clinicians, automated segmentation models are crucial. In this work, we consider the publicly available dataset ATLAS v2.0 to benchmark various end-to-end supervised U-Net style models. Specifically, we have benchmarked models on both 2D and 3D brain images and evaluated them using standard metrics. We have achieved the highest Dice score of 0.583 on the 2D transformer-based model and 0.504 on the 3D residual U-Net respectively. We have conducted the Wilcoxon test for 3D models to correlate the relationship between predicted and actual stroke volume.

Abstract

We present a Spiral Visualization that facilitates users to visually comprehend large graphs. Spiral Visualization is a representation that highlights key aspects of networks, including the number, size, and density of communities, important or central nodes within communities, centrality distribution within communities, connections between communities, and connections between nodes. To facilitate analysis and comprehension of networks using various interaction techniques, such as zooming, tooltip, and highlight, we have implemented a Spiral Visualization dashboard. We conducted a qualitative user study incorporating observation, think-aloud protocols, and participant rating of confidence and easiness to assess the usability and suitability of our visualization. The findings suggest that our visualization is appropriate for the network tasks evaluated. However, tasks requiring color comparison, such as identifying the densest community and comparing community densities were found to be more challenging to perform.

Abstract

With Cardiovascular Diseases on the rise around the world, Electrocardiograms (ECGs) play a crucial role in their diagnosis owing to their non-invasive nature and simplicity. Medical professionals typically use 12-lead ECGs for medical analysis but gathering 12-lead ECG data is an arduous task outside clinical setting. Modern wearables can collect an ECG with fewer leads than the standard 12 leads. However, medical professionals and conventional ECG analysis software find this reduced lead set data challenging to interpret. By using the reduced lead set data to create standard 12-lead ECG data, ECG reconstruction can solve this issue. This paper proposes a novel single-lead to multi-lead ECG reconstruction solution using a modified Attention U-net framework. Using only the lead II of ECG, our model is able to reproduce the other 11 leads of conventional 12-lead ECG with a Pearson correlation, Mean square error and R-squared value of 0.805, 0.0122 and 0.639, respectively. Further, a single combined model is used to reconstruct all 11 leads simultaneously, improving performance and simultaneously reducing the computational resources needed for training compared to current literature in the field. In comparison to previous works, which only reconstruct small ECG segments, our model is trained to reconstruct longer 10-second ECG signals. We demonstrate our model’s ability for real-life utilisation using a cardiovascular disease classification task. A deep learning model was trained for multi-disease classification on actual 12-lead ECG data and was tested on both original and reconstructed 12-lead ECG signals. The classification accuracies for the original and reconstructed signals were comparable, portraying that our reconstruction model can preserve diagnostically relevant artefacts in its reconstructed signals. This work provides a new promising solution in the field of single-lead ECG reconstruction, taking us a step closer to bridging the divide between reduced lead set data and existing 12-lead ECG end users like clinicians and automatic ECG classifiers.