AGAnanya Sen Gupta

Ananya Sen Gupta

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Ananya Sen Gupta has a total of 60 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Quantifying the Relative Weathering of Toxic Hydrocarbons From the Deepwater Horizon Oil Spill Using Peak-Cognizant Raw GC-MS Signal Processing

    by Tonmoy Biswas, Fabian Mullerdahlberg, Bernice Kubicek, Alexandra Zelenski, Lokeshwari Potluri, Allison Flores, Trevor D. Smith, Diego A. Diaz, Ananya Sen Gupta, Laura M. Basirico, et al.

    In this work, we propose a new graph-based data science method to perform peak-cognizant automated analysis of the raw signal from gas chromatography mass spectrometry (GC-MS) data. The application of interest is the environmental weathering of complex mixtures, e.g. crude oil from the Deepwater Horizon oil spill in April 2010. Specifically, we introduce a graph-based representation of the different peaks, target and non-target, in the raw GC-MS signal, which are interpreted as analytes by environmental chemists. Such representation enables us to (i) autonomously interpret the rich complexity of analytes present in the raw GC-MS signal, as well as (ii) autonomously quantify how each analyte degrades across different samples taken from the same source at different times. Accordingly, we present our findings in the form of persistent graphs based on a portfolio of weathered crude oil samples from the Deepwater Horizon 2010 spill, collected across multiple sites across the Gulf of Mexico. The results show considerable variability across different classes of analytes, as expected based on their well-known resilience to environmental weathering. Specifically, we consider analytes such as pyrenes, phenanthrenes, etc. that degrade under environmental weathering processes, and compare their persistence graphs against analytes like hopanes, which are well-known to be recalcitrant against environmental degradation. The value proposition of this work study hydrocarbon persistence autonomously from the raw instrument signal in a peak-cognizant way, which will significantly reduce lab personnel time to manually interpret the weathering profile of these individual analytes. Our work is distinct from statistical chemometric studies which study such analytes in the aggregate, and where such peak-cognizant quantification and interpretation is not possible. In particular, the autonomous peak-cognizant analysis method proposed allows the discovery of persistent non-target analytes in the raw instrument signal, which have never been studied by experts, and buried in the aggregate statistics of chemometric approaches.We also posit the potential application to other complex mixtures, such as soil, and related weathering of soil contaminants, based on this proposed method.

  • Open Access

    Graph-Based Contaminant Identification in Mixture of Analytes Relevant to Soil Studies

    by Tonmoy Biswas, Ananya Sengupta, Patricia Matute, Jessica Stephens, Laura Basirico, Kevin Armbrust

  • Shallow Neural Network Training via Atomic Norms and Semidefinite Programming

    by Andrew J. Christensen, Ananya Sen Gupta

    Neural networks have achieved remarkable results across numerous scientific domains because of their ability to uncover complex patterns. However, despite their effectiveness, these networks rely on heuristic training of highly non-convex objective functions, limiting theoretical understanding and practical reliability. Recent work has shown that shallow neural networks with scalar outputs can be formulated as convex optimization problems, bridging empirical success with theory. In this work, we build upon this framework for vector-valued outputs, introducing a convex formulation for two-layer ReLU networks based on an atomic norm and expressible as a semidefinite program (SDP). This yields a principled convex relaxation of multi-output networks that is both expressive and tractable. We validate the approach using standard SDP solvers, demonstrating its feasibility. These results extend convex neural network training beyond scalar outputs and provide a foundation for scalable, robust alternatives to current heuristic deep learning methods. Our method achieved a 7.3% increase in classification accuracy compared to a baseline convex multi-output network.

  • Evaluating the generalization of complex-weight neural networks over simulated Lamb wave responses from hollow spheres

    by Timothy J. Linhardt, Ananya Sen Gupta, Ivars Kirsteins

    The advancement of complex-valued machine learning brings about new potential for the neglected phase information in acoustics research. A comparison between models that either ignore phase or represent complex numbers as pairs of reals and fully complex numbers yields results that indicate that complex-valued networks are as good as or better than the best fully real option, while having roughly half of the computer memory cost. This is performed using simulated partial wave responses (Lamb waves) for hollow spheres suspended in an infinite homogenous ocean. These spheres of varying thickness are classified based on material with fully connected networks applied to a frequency passband. For the hardest to classify subset of the generated data, there was comparable classification confidence and accuracy observed between the best-performing real network and the complex network. Perfect classification accuracy for unseen partial wave response data was achieved in some trained models, which suggests a disparity in minima in the gradient space and promotes further study into noise augmentation and convolution with simulated multipath channels.


  • Open Access

    Underwater Small Target Classification Using Sparse Multi-View Discriminant Analysis and the Invariant Scattering Transform

    by Andrew Christensen, Ananya Sen Gupta, Ivars Kirsteins

    Sonar automatic target recognition (ATR) systems suffer from complex acoustic scattering, background clutter, and waveguide effects that are ever-present in the ocean. Traditional signal processing techniques often struggle to distinguish targets when noise and complicated target geometries are introduced. Recent advancements in machine learning and wavelet theory offer promising directions for extracting informative features from sonar return data. This work introduces a feature extraction and dimensionality reduction technique using the invariant scattering transform and Sparse Multi-view Discriminant Analysis for identifying highly informative features in the PONDEX09/PONDEX10 datasets. The extracted features are used to train a support vector machine classifier that achieves an average classification accuracy of 97.3% using six unique targets.


  • Open Access

    Enabling in situ validation of mitigation algorithms for magnetic interference via a laboratory-generated dataset

    by Matthew G. Finley, Allison M. Flores, Katherine J. Morris, Robert M. Broadfoot, Sam Hisel, Jason Homann, Chris Piker, Ananya Sen Gupta, David M. Miles

    Magnetometer measurements are one of the critical components necessary for improving our understanding of the intricate physical processes coupling mass, momentum, and energy within near-Earth space and throughout our solar system. However, these measurements are often contaminated by stray magnetic fields from the spacecraft hosting the magnetic-field sensors, and the data often require the application of interference mitigation algorithms prior to scientific use. Rigorous numerical validation of these techniques can be challenging when the techniques are applied to in situ spaceflight data as a ground truth for the local magnetic field is often unavailable. This paper introduces and details the generation of an open-source dataset designed to facilitate the assessment of interference mitigation techniques for magnetic-field data collected during spaceflight missions. The dataset contains over 100 h of magnetic-field data, comprising mixtures of near-direct-current (near-DC) trends, physically synthesized interference, and pseudo-geophysical phenomena. These constituent source signals have been independently captured by four synchronized magnetometers sampling at a high cadence and combined into 30 min intervals of data representing events and interference seen in historic missions. The physical locations of the four magnetometers relative to the interference sources enable researchers to test their interference mitigation algorithms with various magnetometer suite configurations, and the dataset also provides a ground truth for the underlying interference signals, enabling the rigorous quantification of the results of past, present, and future interference mitigation efforts.


  • Open Access

    Comment on egusphere-2024-87

    by Matthew G. Finley, Allison M. Flores, Katherine J. Morris, Robert M. Broadfoot, Sam Hisel, Jason Homann, Chris Piker, Ananya Sen Gupta, David M. Miles

    Magnetometer measurements are one of the critical components necessary to improve our understanding of the intricate physical processes coupling mass, momentum, and energy within near-Earth space and throughout our solar system. However, these measurements are often contaminated by stray magnetic fields from the spacecraft hosting the magnetic field sensors, and the data often requires the application of interference mitigation algorithms prior to scientific use. Rigorous numerical validation of these techniques can be challenging when they are applied to in-situ spaceflight data, as a ground truth for the local magnetic field is often unavailable. This manuscript introduces and details the generation of an open-source dataset designed to facilitate the assessment of interference mitigation techniques for magnetic field data collected during spaceflight missions. The dataset contains over 100 hours of magnetic field data comprising mixtures of near-DC trends, physically-synthesized interference, and pseudo-geophysical phenomena. These constituent source signals have been independently captured by four synchronized magnetometers sampling at high cadence and combined into 30-minute intervals of data representative of events and interference seen in historic missions. The physical location of the four magnetometers relative to the interference sources enables researchers to test their interference mitigation algorithms with various magnetometer suite configurations, and the dataset also provides a ground truth for the underlying interference signals, enabling rigorous quantification of the results of past, present, and future interference mitigation efforts.

  • Open Access

    Comment on egusphere-2024-87

    by Matthew G. Finley, Allison M. Flores, Katherine J. Morris, Robert M. Broadfoot, Sam Hisel, Jason Homann, Chris Piker, Ananya Sen Gupta, David M. Miles

    Magnetometer measurements are one of the critical components necessary to improve our understanding of the intricate physical processes coupling mass, momentum, and energy within near-Earth space and throughout our solar system. However, these measurements are often contaminated by stray magnetic fields from the spacecraft hosting the magnetic field sensors, and the data often requires the application of interference mitigation algorithms prior to scientific use. Rigorous numerical validation of these techniques can be challenging when they are applied to in-situ spaceflight data, as a ground truth for the local magnetic field is often unavailable. This manuscript introduces and details the generation of an open-source dataset designed to facilitate the assessment of interference mitigation techniques for magnetic field data collected during spaceflight missions. The dataset contains over 100 hours of magnetic field data comprising mixtures of near-DC trends, physically-synthesized interference, and pseudo-geophysical phenomena. These constituent source signals have been independently captured by four synchronized magnetometers sampling at high cadence and combined into 30-minute intervals of data representative of events and interference seen in historic missions. The physical location of the four magnetometers relative to the interference sources enables researchers to test their interference mitigation algorithms with various magnetometer suite configurations, and the dataset also provides a ground truth for the underlying interference signals, enabling rigorous quantification of the results of past, present, and future interference mitigation efforts.

  • Open Access

    Comment on egusphere-2024-87

    by Matthew G. Finley, Allison M. Flores, Katherine J. Morris, Robert M. Broadfoot, Sam Hisel, Jason Homann, Chris Piker, Ananya Sen Gupta, David M. Miles

    Magnetometer measurements are one of the critical components necessary to improve our understanding of the intricate physical processes coupling mass, momentum, and energy within near-Earth space and throughout our solar system. However, these measurements are often contaminated by stray magnetic fields from the spacecraft hosting the magnetic field sensors, and the data often requires the application of interference mitigation algorithms prior to scientific use. Rigorous numerical validation of these techniques can be challenging when they are applied to in-situ spaceflight data, as a ground truth for the local magnetic field is often unavailable. This manuscript introduces and details the generation of an open-source dataset designed to facilitate the assessment of interference mitigation techniques for magnetic field data collected during spaceflight missions. The dataset contains over 100 hours of magnetic field data comprising mixtures of near-DC trends, physically-synthesized interference, and pseudo-geophysical phenomena. These constituent source signals have been independently captured by four synchronized magnetometers sampling at high cadence and combined into 30-minute intervals of data representative of events and interference seen in historic missions. The physical location of the four magnetometers relative to the interference sources enables researchers to test their interference mitigation algorithms with various magnetometer suite configurations, and the dataset also provides a ground truth for the underlying interference signals, enabling rigorous quantification of the results of past, present, and future interference mitigation efforts.

  • Open Access

    Artificial data augmentation using braided human features for underwater acoustics

    by Ananya Sen Gupta, Andrew J. Christensen, Anjali Mathews, Subhajit Das, Ivars P. Kirsteins

    Underwater acoustics datasets provide rich environmental information that can potentially be mined using popular machine learning architectures. However, such knowledge discovery is typically limited by the high feature uncertainty, lack of robust ground truths, limited availability of public-domain data, and difficulty of reproducing experimental conditions in field experiments. This poses a data science challenge to study persistent target features, as they can only be learnt and classified robustly if large scale robust training and testing datasets are available. This “nano-data” problem can be potentially solved by augmenting the training and testing data repositories with a two-pronged approach: (i) physics-driven simulations of desired features, which offer interpretable ground truths but typically cannot emulate practical open-sea experiments; (ii) geometric-proxy augmentation using creatively constructed non-domain datasets that emulate the desired feature geometries and environmental effects. For example, a skilled dancer can emulate specific features geometries through geometric contortions of the human body. Such choreographed dance movements can be reproduced reliably against different types of stage environments, structured and unstructured, to emulate the oceanic environment under different conditions. We will present preliminary results in data augmentation from both approaches and discuss the trade-offs between physics-oriented simulations and geometry-driven proxy augmentation. [ONR grant N000142312503.]