arXiv subjects

Indranil Dutta

Publications and source records attributed to Indranil Dutta.

10 recordsLinked to original sources

arXiv2026

Figurative Justice: Detecting metaphors in Hindi judgements with qualitative assessment and transformers

Metaphors are figurative use of words for conceptual mapping. Metaphor detection in the legal context has been crucial as metaphors are persuasive juridical means of creating legal meaning and concepts resulting in significant consequences. Metaphorical framing in legal discourse by judges, lawyers, and legislators brings about real-time implications upon individuals and influences judicial decision-making, argumentation and interpretation of laws. This is crucial in Human Rights infringement cases where language determines severity of punishment, public perception and judicial outcomes. While automatic metaphor detection in major languages like English, Spanish, Polish, Lithuanian have aided in understanding inherent intentions of metaphorical use of language, there is no such attempt in low-resource languages like Hindi. The dearth of annotated legal corpora in Hindi makes it difficult to develop NLP models and detect metaphors in judicial proceedings. In the Indian context, Convolutional Neural Networks (CNNs) have been used for classification of bail judgements, however there are no existing models designed for metaphor detection. We present a Hindi Legal Metaphor Corpus (HiLeMe) by isolating judgements from Hindi Legal Data Corpus (HLDC). Legal experts annotated HiLeMe to classify metaphorical constructions using the MIPVU schema. We downstreamed an mBERT on Hindi legal metaphor detection task. We built a transformer-based architecture for metaphor detection that are known to outperform traditional models in legal classification tasks. This model provides insights into the judicial psyche for decoding judicial decisions. Our research contributes to advancing automated models in legal discourse in low-resource languages like Hindi and envisages adoption into 22 Indian schedule languages.

cs.CL↗

arXiv2026

Multi-Resolution Feature Stem for Diabetic Retinopathy lesion segmentation

Diabetic Retinopathy (DR) is a leading cause of preventable blindness worldwide, requiring automated lesion segmentation using deep learning models for early detection and monitoring. However, DR lesions vary dramatically in size from tiny microaneurysms to large hemorrhages and exudates. This variability creates conflicting demands on the model architecture and input resolution, posing a challenge for effective design. This work investigates the impact of input resolution on different lesion types. Through systematic experimentation with multiple architectures (U-Net, UNet++, Vision Transformers, DeepLabV3+) at $512 \times 512$ and $1024 \times 1024$ resolutions, we identify a critical, counter-intuitive phenomenon where increasing input resolution has opposing effects on different lesion types. We demonstrate that while higher resolution is essential for resolving fine-grained microaneurysms, it can unexpectedly degrade performance on larger hemorrhages. This finding challenges the common assumption that higher resolution is uniformly beneficial. To address this, we propose a novel Multi-Resolution Feature Stem, an input-level pyramid integrated with a UNet++ backbone. This architecture processes multiple scales in parallel, capturing fine-grained details without sacrificing contextual information. This work contributes crucial empirical evidence of this complex, resolution-dependent behavior and a practical, parameter-efficient architecture that successfully resolves this trade-off.

cs.CV↗

arXiv2025

Quantum-assured magnetic navigation achieves positioning accuracy better than a strategic-grade INS in airborne and ground-based field trials

Modern navigation systems rely critically on GNSS, which in many cases is unavailable or unreliable (e.g. due to jamming or spoofing). For this reason there is great interest in augmenting backup navigation systems such as inertial navigation systems (INS) with additional modalities that reduce positioning error in the absence of reliable GNSS. Magnetic-anomaly navigation is one such approach, providing passive, non-jammable navigation through periodic position fixes obtained by comparing local measurements of Earth's crustal field against known anomaly maps. Despite its potential, existing MagNav efforts have been limited by magnetometer performance and platform noise; solutions addressing these problems have proven either too brittle or impractical for realistic deployment. Here we demonstrate a quantum-assured MagNav solution based on proprietary quantum magnetometers with by a novel denoising and map-matching algorithms. The system fits on fixed-wing drones or in the avionics bay of a commercial airliner. We present trials at altitudes up to 19000 feet, testing onboard and outboard quantum magnetometers comparing against a strategic-grade INS. Our MagNav solution achieves superior performance, delivering up to 46x better positioning error than the velocity-aided INS; the best final positioning accuracy we achieve is 22m or 0.006% of the flight distance. Airborne trials consistently achieve at least 11x advantage over the INS across varying conditions, altitudes, and flight patterns. The system learns model parameters online without special vehicle maneuvers providing robustness to various configuration changes (e.g. changing payload or latitude). Our trials also include the first successful MagNav performed in a ground vehicle using publicly-available anomaly maps, delivering bounded positioning error 7x lower than the INS, with both systems in strapdown configuration.

quant-ph↗

arXiv2024

Convolutional Neural Networks can achieve binary bail judgement classification

There is an evident lack of implementation of Machine Learning (ML) in the legal domain in India, and any research that does take place in this domain is usually based on data from the higher courts of law and works with English data. The lower courts and data from the different regional languages of India are often overlooked. In this paper, we deploy a Convolutional Neural Network (CNN) architecture on a corpus of Hindi legal documents. We perform a bail Prediction task with the help of a CNN model and achieve an overall accuracy of 93\% which is an improvement on the benchmark accuracy, set by Kapoor et al. (2022), albeit in data from 20 districts of the Indian state of Uttar Pradesh.

cs.CL↗

arXiv2024

Attentive Fusion: A Transformer-based Approach to Multimodal Hate Speech Detection

With the recent surge and exponential growth of social media usage, scrutinizing social media content for the presence of any hateful content is of utmost importance. Researchers have been diligently working since the past decade on distinguishing between content that promotes hatred and content that does not. Traditionally, the main focus has been on analyzing textual content. However, recent research attempts have also commenced into the identification of audio-based content. Nevertheless, studies have shown that relying solely on audio or text-based content may be ineffective, as recent upsurge indicates that individuals often employ sarcasm in their speech and writing. To overcome these challenges, we present an approach to identify whether a speech promotes hate or not utilizing both audio and textual representations. Our methodology is based on the Transformer framework that incorporates both audio and text sampling, accompanied by our very own layer called "Attentive Fusion". The results of our study surpassed previous state-of-the-art techniques, achieving an impressive macro F1 score of 0.927 on the Test Set.

cs.CL↗

arXiv2023

Is Attention always needed? A Case Study on Language Identification from Speech

Language Identification (LID) is a crucial preliminary process in the field of Automatic Speech Recognition (ASR) that involves the identification of a spoken language from audio samples. Contemporary systems that can process speech in multiple languages require users to expressly designate one or more languages prior to utilization. The LID task assumes a significant role in scenarios where ASR systems are unable to comprehend the spoken language in multilingual settings, leading to unsuccessful speech recognition outcomes. The present study introduces convolutional recurrent neural network (CRNN) based LID, designed to operate on the Mel-frequency Cepstral Coefficient (MFCC) characteristics of audio samples. Furthermore, we replicate certain state-of-the-art methodologies, specifically the Convolutional Neural Network (CNN) and Attention-based Convolutional Recurrent Neural Network (CRNN with attention), and conduct a comparative analysis with our CRNN-based approach. We conducted comprehensive evaluations on thirteen distinct Indian languages and our model resulted in over 98\% classification accuracy. The LID model exhibits high-performance levels ranging from 97% to 100% for languages that are linguistically similar. The proposed LID model exhibits a high degree of extensibility to additional languages and demonstrates a strong resistance to noise, achieving 91.2% accuracy in a noisy setting when applied to a European Language (EU) dataset.

cs.LG↗

arXiv2018

Experiment to detect dark energy forces using atom interferometry

The accelerated expansion of the universe motivates a wide class of scalar field theories that modify gravity on large scales. In regions where the weak field limit of General Relativity has been confirmed by experiment, such theories need a screening mechanism to suppress the new force. We have measured the acceleration of an atom toward a macroscopic test mass inside a high vacuum chamber, where the new force is unscreened in some theories. Our measurement, made using atom interferometry, shows that the attraction between atoms and the test mass does not differ appreciably from Newtonian gravity. This result places stringent limits on the free parameters in chameleon and symmetron theories of modified gravity.

physics.atom-ph↗

arXiv2016

Cold-atom Inertial Sensor without Deadtime

We report the operation of a cold-atom inertial sensor in a joint interrogation scheme, where we simultaneously prepare a cold-atom source and operate an atom interferometer in order to eliminate dead times. Noise aliasing and dead times are consequences of the sequential operation which is intrinsic to cold-atom atom interferometers. Both phenomena have deleterious effects on the performance of these sensors. We show that our continuous operation improves the short-term sensitivity of atom interferometers, by demonstrating a record rotation sensitivity of $100$ nrad.s$^{-1}/\sqrt{\rm Hz}$ in a cold-atom gyroscope of $11$ cm$^2$ Sagnac area. We also demonstrate a rotation stability of $1$ nrad.s$^{-1}$ after $10^4$ s of integration, improving previous results by an order of magnitude. We expect that the continuous operation will allow cold-atom inertial sensors with long interrogation time to reach their full sensitivity, determined by the quantum noise limit.

physics.atom-ph↗

arXiv2016

Metrology with Atom Interferometry: Inertial Sensors from Laboratory to Field Applications

Developments in atom interferometry have led to atomic inertial sensors with extremely high sensitivity. Their performances are for the moment limited by the ground vibrations, the impact of which is exacerbated by the sequential operation, resulting in aliasing and dead time. We discuss several experiments performed at LNE-SYRTE in order to reduce these problems and achieve the intrinsic limit of atomic inertial sensors. These techniques have resulted in transportable and high-performance instruments that participate in gravity measurements, and pave the way to applications in inertial navigation.

physics.atom-ph↗

arXiv2014

The Sagnac effect: 20 years of development in matter-wave interferometry

Since the first atom interferometry experiments in 1991, measurements of rotation through the Sagnac effect in open-area atom interferometers has been studied. These studies have demonstrated very high sensitivity which can compete with state-of-the-art optical Sagnac interferometers. Since the early 2000s, these developments have been motivated by possible applications in inertial guidance and geophysics. Most matter-wave interferometers that have been investigated since then are based on two-photon Raman transitions for the manipulation of atomic wave packets. Results from the two most studied configurations, a space-domain interferometer with atomic beams and a time-domain interferometer with cold atoms, are presented and compared. Finally, the latest generation of cold atom interferometers and their preliminary results are presented.

physics.atom-ph↗