Seshan Srirangarajan

Aman Verma, Seshan Srirangarajan, Sumantra Dutta Roy

Quantifying biometric characteristics within hand gestures involve derivation of fitness scores from a gesture and identity aware feature space. However, evaluating the quality of these scores remains an open question. Existing biometric capacity estimation literature relies upon error rates. But these rates do not indicate goodness of scores. Thus, in this manuscript we present an exhaustive set of evaluation measures. We firstly identify ranking order and relevance of output scores as the primary basis for evaluation. In particular, we consider both rank deviation as well as rewards for: (i) higher scores of high ranked gestures and (ii) lower scores of low ranked gestures. We also compensate for correspondence between trends of output and ground truth scores. Finally, we account for disentanglement between identity features of gestures as a discounting factor. Integrating these elements with adequate weighting, we formulate advanced acceptance score as a holistic evaluation measure. To assess effectivity of the proposed we perform in-depth experimentation over three datasets with five state-of-the-art (SOTA) models. Results show that the optimal score selected with our measure is more appropriate than existing other measures. Also, our proposed measure depicts correlation with existing measures. This further validates its reliability. We have made our \href{https://github.com/AmanVerma2307/MeasureSuite}{code} public.

Prasannaa Kumar D., Gulshan Kumar, Jay Dhariwal et al.

The rising demand for low-cost air quality monitors stems from increased public awareness and interest within the research community. These monitors play a pivotal role in empowering citizens and scientists to comprehend spatiotemporal variations in air quality parameters, aiding in the formulation of effective mitigation policies. The primary challenge lies in the diverse array of application scenarios these monitors encounter. The developed data logging device is exceptionally well-suited for air quality monitoring applications, offering exceptional versatility by seamlessly operating on a range of power sources, including solar energy, batteries, and direct electrical supply. The integration of a built-in battery charger enhances its applicability for deployment in regions with solar power or intermittent electricity availability. To ensure strong network connectivity, the advanced datalogger seamlessly integrates with WiFi, Bluetooth, and LoRaWAN networks. A notable feature is its adaptable MCU system, enabling users to swap the MCU based on specific connectivity, power, and computational requirements. Importantly, the system carefully identifies key parameters crucial for both indoor and outdoor air quality assessment, customizing sensor selection accordingly. Furthermore, optimization efforts have prioritized energy efficiency, enabling the system to function with minimal power consumption while maintaining data integrity. Additional I2C and UART ports facilitate the monitoring of supplementary parameters.

Vishnuvardhan Purma, Suhas Srinath, Seshan Srirangarajan et al.

Histopathological image segmentation is a laborious and time-intensive task, often requiring analysis from experienced pathologists for accurate examinations. To reduce this burden, supervised machine-learning approaches have been adopted using large-scale annotated datasets for histopathological image analysis. However, in several scenarios, the availability of large-scale annotated data is a bottleneck while training such models. Self-supervised learning (SSL) is an alternative paradigm that provides some respite by constructing models utilizing only the unannotated data which is often abundant. The basic idea of SSL is to train a network to perform one or many pseudo or pretext tasks on unannotated data and use it subsequently as the basis for a variety of downstream tasks. It is seen that the success of SSL depends critically on the considered pretext task. While there have been many efforts in designing pretext tasks for classification problems, there haven't been many attempts on SSL for histopathological segmentation. Motivated by this, we propose an SSL approach for segmenting histopathological images via generative diffusion models in this paper. Our method is based on the observation that diffusion models effectively solve an image-to-image translation task akin to a segmentation task. Hence, we propose generative diffusion as the pretext task for histopathological image segmentation. We also propose a multi-loss function-based fine-tuning for the downstream task. We validate our method using several metrics on two publically available datasets along with a newly proposed head and neck (HN) cancer dataset containing hematoxylin and eosin (H\&E) stained images along with annotations. Codes will be made public at https://github.com/suhas-srinath/GenSelfDiff-HIS.

Bodhibrata Mukhopadhyay, Seshan Srirangarajan, Subrat Kar

The ability of a sensor node to determine its location in a sensor network is important in many applications. The infrastructure for the location-based services is an easy target for malicious attacks. We address scenarios where malicious node(s) attempt to disrupt, in an uncoordinated or coordinated manner, the localization process of a target node. We propose four techniques for secure localization: weighted least square (WLS), secure weighted least square (SWLS), and $\ell_1$-norm based techniques LN-1 and LN-1E, in a network that includes one or more compromised anchor nodes. WLS and SWLS techniques are shown to offer significant advantage over existing techniques by assigning larger weights to the anchor nodes that are closer to the target node, and by detecting the malicious nodes and eliminating their measurements from the localization process. In a coordinated attack, the localization problem can be posed as a plane fitting problem where the measurements from non-malicious and malicious anchor nodes lie on two different planes. LN-1E technique estimates these two planes and prevents disruption of the localization process. The Cramer-Rao lower bound (CRLB) for the position estimate is also derived. The proposed techniques are shown to provide better localization accuracy than the existing algorithms.