arXiv subjects

Soumya Ghosh

Publications and source records attributed to Soumya Ghosh.

At least 37 records · Page 2Linked to original sources

arXiv2024

Are you using test log-likelihood correctly?

Test log-likelihood is commonly used to compare different models of the same data or different approximate inference algorithms for fitting the same probabilistic model. We present simple examples demonstrating how comparisons based on test log-likelihood can contradict comparisons according to other objectives. Specifically, our examples show that (i) approximate Bayesian inference algorithms that attain higher test log-likelihoods need not also yield more accurate posterior approximations and (ii) conclusions about forecast accuracy based on test log-likelihood comparisons may not agree with conclusions based on root mean squared error.

stat.ML↗

arXiv2023

Assessment of Prediction Intervals Using Uncertainty Characteristics Curves

Accurate quantification of model uncertainty has long been recognized as a fundamental requirement for trusted AI. In regression tasks, uncertainty is typically quantified using prediction intervals calibrated to an ad-hoc operating point, making evaluation and comparison across different studies relatively difficult. Our work leverages: (1) the concept of operating characteristics curves and (2) the notion of a gain over a null reference, to derive a novel operating point agnostic assessment methodology for prediction intervals. The paper defines the Uncertainty Characteristics Curve and demonstrates its utility in selected scenarios. We argue that the proposed method addresses the current need for comprehensive assessment of prediction intervals and thus represents a valuable addition to the uncertainty quantification toolbox.

cs.LG↗

arXiv2023

Photochargeable Li-ion Battery: Type II Heterojunction Exposes Underlying Band Gap in 'Metallic' Lithiated MoS2

Light chargeable metal-ion batteries using semiconductor heterostructures are gaining enormous interest. A few such heterostructures such as MoS2/MoOy and TiS2/TiO2 have been shown to function as photocathodes in photochargeable Li-ion batteries, where the type II set-up has been proposed to generate spatially separated (longer-lived) excitons upon photo-exposure. The Li intercalated MoS2, generated during the discharge cycle of the battery, undergoes a phase transition from the semiconducting (2H) to a metallic (1T) phase, in contrast to its TiS2 counterpart, casting a doubt over the photocharging process. Here, employing density functional theory based traditional as well as unconventional computational schemes along with relevant spectroscopic techniques, we show that in LixMoS2/MoO3 heterostructure an underlying band gap of LixMoS2 is exposed, upto a certain value of x, due to dispersion of electron density onto MoO3 justifying the observed photocharging. We believe that the general concepts explored in this study will be important in the rational design of photo-cathode materials in Li-ion batteries.

cond-mat.mtrl-sci↗

arXiv2023

Quantifying Defects and Finite Size Effects in Graphene Oxide Models

Oxidation of graphite and subsequent exfoliation leads to single layer graphene oxide (GO). GO has found many applications across diverse fields including medicinal chemistry, catalysis as well as a precursor for graphene. One of the key structural features of GO is the presence of different kinds of defects. Molecular dynamics simulations with ReaxFF force fields have been widely used to model realistic representations of GO that include defects of various types. In these simulations, one can vary the extent and distribution of the defects by changing the initial O/C ratio. It is therefore very important to employ a proper measure of the defect density. Traditionally, the total number of non-graphitic carbon atoms have been employed to quantify the amount of defects. Our simulations suggest that this parameter may not be a good measure at low defect densities. Herein, we introduce a hitherto unexplored metric, relative area of the defects, to gauge the defect density. We show that this metric has desirable properties at both low and high defect densities. Additionally, we investigate the changes in the defect distribution and mechanical properties upon varying the size of the simulation cell.

cond-mat.mtrl-sci↗

arXiv2023

Reliable Gradient-free and Likelihood-free Prompt Tuning

Due to privacy or commercial constraints, large pre-trained language models (PLMs) are often offered as black-box APIs. Fine-tuning such models to downstream tasks is challenging because one can neither access the model's internal representations nor propagate gradients through it. This paper addresses these challenges by developing techniques for adapting PLMs with only API access. Building on recent work on soft prompt tuning, we develop methods to tune the soft prompts without requiring gradient computation. Further, we develop extensions that in addition to not requiring gradients also do not need to access any internal representation of the PLM beyond the input embeddings. Moreover, instead of learning a single prompt, our methods learn a distribution over prompts allowing us to quantify predictive uncertainty. Ours is the first work to consider uncertainty in prompts when only having API access to the PLM. Finally, through extensive experiments, we carefully vet the proposed methods and find them competitive with (and sometimes even improving on) gradient-based approaches with full access to the PLM.

cs.LG↗

arXiv2023

Photo-Rechargeable Li Ion Batteries using TiS2 Cathode

Photo-rechargeable (solar) battery can be considered as an energy harvesting cum storage system, where it can charge the conventional metal-ion battery using light instead of electricity, without having other parasitic reactions. Here we demonstrate a two-electrode lithium ion solar battery with multifaceted TiS2-TiO2 hybrid sheets as cathode. Choice of TiS2-TiO2 electrode ensures the formation of a type II semiconductor heterostructure while the lateral heterostructure geometry ensures high mass/charge transfer and light interactions with the electrode. TiS2 has a higher lithium binding energy (1.6 eV) than TiO2 (1.03 eV), ensuring the possibilities of higher amount of Li ion insertion to TiS2 and hence the maximum recovery with the photocharging, as further confirmed by the experiments. Apart from the demonstration of solar solid-state batteries, the charging of lithium ion full cell with light indicates the formation of lithium intercalated graphite compounds, ensuring the charging of the battery without any other parasitic reactions at the electrolyte or electrode-electrolyte interfaces. Possible mechanisms proposed here for the charging and discharging processes of solar batteries, based on our experimental and theoretical results, indicate the potential of such systems in forthcoming era of renewable energies.

physics.app-ph↗

arXiv2022

Post-hoc Uncertainty Learning using a Dirichlet Meta-Model

It is known that neural networks have the problem of being over-confident when directly using the output label distribution to generate uncertainty measures. Existing methods mainly resolve this issue by retraining the entire model to impose the uncertainty quantification capability so that the learned model can achieve desired performance in accuracy and uncertainty prediction simultaneously. However, training the model from scratch is computationally expensive and may not be feasible in many situations. In this work, we consider a more practical post-hoc uncertainty learning setting, where a well-trained base model is given, and we focus on the uncertainty quantification task at the second stage of training. We propose a novel Bayesian meta-model to augment pre-trained models with better uncertainty quantification abilities, which is effective and computationally efficient. Our proposed method requires no additional training data and is flexible enough to quantify different uncertainties and easily adapt to different application settings, including out-of-domain data detection, misclassification detection, and trustworthy transfer learning. We demonstrate our proposed meta-model approach's flexibility and superior empirical performance on these applications over multiple representative image classification benchmarks.

cs.LG↗

arXiv2022

Fair Infinitesimal Jackknife: Mitigating the Influence of Biased Training Data Points Without Refitting

In consequential decision-making applications, mitigating unwanted biases in machine learning models that yield systematic disadvantage to members of groups delineated by sensitive attributes such as race and gender is one key intervention to strive for equity. Focusing on demographic parity and equality of opportunity, in this paper we propose an algorithm that improves the fairness of a pre-trained classifier by simply dropping carefully selected training data points. We select instances based on their influence on the fairness metric of interest, computed using an infinitesimal jackknife-based approach. The dropping of training points is done in principle, but in practice does not require the model to be refit. Crucially, we find that such an intervention does not substantially reduce the predictive performance of the model but drastically improves the fairness metric. Through careful experiments, we evaluate the effectiveness of the proposed approach on diverse tasks and find that it consistently improves upon existing alternatives.

cs.LG↗

arXiv2022

Measuring the robustness of Gaussian processes to kernel choice

Gaussian processes (GPs) are used to make medical and scientific decisions, including in cardiac care and monitoring of atmospheric carbon dioxide levels. Notably, the choice of GP kernel is often somewhat arbitrary. In particular, uncountably many kernels typically align with qualitative prior knowledge (e.g.\ function smoothness or stationarity). But in practice, data analysts choose among a handful of convenient standard kernels (e.g.\ squared exponential). In the present work, we ask: Would decisions made with a GP differ under other, qualitatively interchangeable kernels? We show how to answer this question by solving a constrained optimization problem over a finite-dimensional space. We can then use standard optimizers to identify substantive changes in relevant decisions made with a GP. We demonstrate in both synthetic and real-world examples that decisions made with a GP can exhibit non-robustness to kernel choice, even when prior draws are qualitatively interchangeable to a user.

stat.ML↗

arXiv2021

Spectral control of nonclassical light using an integrated thin-film lithium niobate modulator

Manipulating the frequency and bandwidth of nonclassical light is essential for implementing frequency-encoded/multiplexed quantum computation, communication, and networking protocols, and for bridging spectral mismatch among various quantum systems. However, quantum spectral control requires a strong nonlinearity mediated by light, microwave, or acoustics, which is challenging to realize with high efficiency, low noise, and on an integrated chip. Here, we demonstrate both frequency shifting and bandwidth compression of nonclassical light using an integrated thin-film lithium niobate (TFLN) phase modulator. We achieve record-high electro-optic frequency shearing of telecom single photons over terahertz range ($\pm$ 641 GHz or $\pm$ 5.2 nm), enabling high visibility quantum interference between frequency-nondegenerate photon pairs. We further operate the modulator as a time lens and demonstrate over eighteen-fold (6.55 nm to 0.35 nm) bandwidth compression of single photons. Our results showcase the viability and promise of on-chip quantum spectral control for scalable photonic quantum information processing.

physics.optics↗

arXiv2021

Mobility driven Cloud-Fog-Edge Framework for Location-aware Services: A Comprehensive Review

With the pervasiveness of IoT devices, smart-phones and improvement of location-tracking technologies huge volume of heterogeneous geo-tagged (location specific) data is generated which facilitates several location-aware services. The analytics with this spatio-temporal (having location and time dimensions) datasets provide varied important services such as, smart transportation, emergency services (health-care, national defence or urban planning). While cloud paradigm is suitable for the capability of storage and computation, the major bottleneck is network connectivity loss. In time-critical application, where real-time response is required for emergency service-provisioning, such connectivity issues increases the latency and thus affects the overall quality of system (QoS). To overcome the issue, fog/ edge topology has emerged, where partial computation is carried out in the edge of the network to reduce the delay in communication. Such fog/ edge based system complements the cloud technology and extends the features of the system. This chapter discusses cloud-fog-edge based hierarchical collaborative framework, where several components are deployed to improve the QoS. On the other side. mobility is another critical factor to enhance the efficacy of such location-aware service provisioning. Therefore, this chapter discusses the concerns and challenges associated with mobility-driven cloud-fog-edge based framework to provide several location-aware services to the end-users efficiently.

cs.DC↗

arXiv2021

Post-hoc loss-calibration for Bayesian neural networks

Bayesian decision theory provides an elegant framework for acting optimally under uncertainty when tractable posterior distributions are available. Modern Bayesian models, however, typically involve intractable posteriors that are approximated with, potentially crude, surrogates. This difficulty has engendered loss-calibrated techniques that aim to learn posterior approximations that favor high-utility decisions. In this paper, focusing on Bayesian neural networks, we develop methods for correcting approximate posterior predictive distributions encouraging them to prefer high-utility decisions. In contrast to previous work, our approach is agnostic to the choice of the approximate inference algorithm, allows for efficient test time decision making through amortization, and empirically produces higher quality decisions. We demonstrate the effectiveness of our approach through controlled experiments spanning a diversity of tasks and datasets.

cs.LG↗

arXiv2021

Uncertainty Quantification 360: A Holistic Toolkit for Quantifying and Communicating the Uncertainty of AI

In this paper, we describe an open source Python toolkit named Uncertainty Quantification 360 (UQ360) for the uncertainty quantification of AI models. The goal of this toolkit is twofold: first, to provide a broad range of capabilities to streamline as well as foster the common practices of quantifying, evaluating, improving, and communicating uncertainty in the AI application development lifecycle; second, to encourage further exploration of UQ's connections to other pillars of trustworthy AI such as fairness and transparency through the dissemination of latest research and education materials. Beyond the Python package (\url{https://github.com/IBM/UQ360}), we have developed an interactive experience (\url{http://uq360.mybluemix.net}) and guidance materials as educational tools to aid researchers and developers in producing and communicating high-quality uncertainties in an effective manner.

cs.AI↗

arXiv2021

Uncertainty Characteristics Curves: A Systematic Assessment of Prediction Intervals

Accurate quantification of model uncertainty has long been recognized as a fundamental requirement for trusted AI. In regression tasks, uncertainty is typically quantified using prediction intervals calibrated to a specific operating point, making evaluation and comparison across different studies difficult. Our work leverages: (1) the concept of operating characteristics curves and (2) the notion of a gain over a simple reference, to derive a novel operating point agnostic assessment methodology for prediction intervals. The paper describes the corresponding algorithm, provides a theoretical analysis, and demonstrates its utility in multiple scenarios. We argue that the proposed method addresses the current need for comprehensive assessment of prediction intervals and thus represents a valuable addition to the uncertainty quantification toolbox.

cs.LG↗

arXiv2021

Integrated photonics on thin-film lithium niobate

Lithium niobate (LN), an outstanding and versatile material, has influenced our daily life for decades: from enabling high-speed optical communications that form the backbone of the Internet to realizing radio-frequency filtering used in our cell phones. This half-century-old material is currently embracing a revolution in thin-film LN integrated photonics. The success of manufacturing wafer-scale, high-quality, thin films of LN on insulator (LNOI), accompanied with breakthroughs in nanofabrication techniques, have made high-performance integrated nanophotonic components possible. With rapid development in the past few years, some of these thin-film LN devices, such as optical modulators and nonlinear wavelength converters, have already outperformed their legacy counterparts realized in bulk LN crystals. Furthermore, the nanophotonic integration enabled ultra-low-loss resonators in LN, which unlocked many novel applications such as optical frequency combs and quantum transducers. In this Review, we cover -- from basic principles to the state of the art -- the diverse aspects of integrated thin-film LN photonics, including the materials, basic passive components, and various active devices based on electro-optics, all-optical nonlinearities, and acousto-optics. We also identify challenges that this platform is currently facing and point out future opportunities. The field of integrated LNOI photonics is advancing rapidly and poised to make critical impacts on a broad range of applications in communication, signal processing, and quantum information.

physics.optics↗

arXiv2020

EVA: Generating Longitudinal Electronic Health Records Using Conditional Variational Autoencoders

Researchers require timely access to real-world longitudinal electronic health records (EHR) to develop, test, validate, and implement machine learning solutions that improve the quality and efficiency of healthcare. In contrast, health systems value deeply patient privacy and data security. De-identified EHRs do not adequately address the needs of health systems, as de-identified data are susceptible to re-identification and its volume is also limited. Synthetic EHRs offer a potential solution. In this paper, we propose EHR Variational Autoencoder (EVA) for synthesizing sequences of discrete EHR encounters (e.g., clinical visits) and encounter features (e.g., diagnoses, medications, procedures). We illustrate that EVA can produce realistic EHR sequences, account for individual differences among patients, and can be conditioned on specific disease conditions, thus enabling disease-specific studies. We design efficient, accurate inference algorithms by combining stochastic gradient Markov Chain Monte Carlo with amortized variational inference. We assess the utility of the methods on large real-world EHR repositories containing over 250, 000 patients. Our experiments, which include user studies with knowledgeable clinicians, indicate the generated EHR sequences are realistic. We confirmed the performance of predictive models trained on the synthetic data are similar with those trained on real EHRs. Additionally, our findings indicate that augmenting real data with synthetic EHRs results in the best predictive performance - improving the best baseline by as much as 8% in top-20 recall.

cs.LG↗

arXiv2020

Approximate Cross-Validation for Structured Models

Many modern data analyses benefit from explicitly modeling dependence structure in data -- such as measurements across time or space, ordered words in a sentence, or genes in a genome. A gold standard evaluation technique is structured cross-validation (CV), which leaves out some data subset (such as data within a time interval or data in a geographic region) in each fold. But CV here can be prohibitively slow due to the need to re-run already-expensive learning algorithms many times. Previous work has shown approximate cross-validation (ACV) methods provide a fast and provably accurate alternative in the setting of empirical risk minimization. But this existing ACV work is restricted to simpler models by the assumptions that (i) data across CV folds are independent and (ii) an exact initial model fit is available. In structured data analyses, both these assumptions are often untrue. In the present work, we address (i) by extending ACV to CV schemes with dependence structure between the folds. To address (ii), we verify -- both theoretically and empirically -- that ACV quality deteriorates smoothly with noise in the initial fit. We demonstrate the accuracy and computational benefits of our proposed methods on a diverse set of real-world applications.

stat.ML↗

arXiv2020

Model Fusion with Kullback--Leibler Divergence

We propose a method to fuse posterior distributions learned from heterogeneous datasets. Our algorithm relies on a mean field assumption for both the fused model and the individual dataset posteriors and proceeds using a simple assign-and-average approach. The components of the dataset posteriors are assigned to the proposed global model components by solving a regularized variant of the assignment problem. The global components are then updated based on these assignments by their mean under a KL divergence. For exponential family variational distributions, our formulation leads to an efficient non-parametric algorithm for computing the fused model. Our algorithm is easy to describe and implement, efficient, and competitive with state-of-the-art on motion capture analysis, topic modeling, and federated learning of Bayesian neural networks.

cs.LG↗