Coreset selection reduces training cost by constructing compact, representative subsets, but existing methods largely assume balanced class distributions. Under imbalance, this assumption yields biased subsets that discard critical minority samples and degrade accuracy. We propose Equitable Coreset Selection (ECS), a framework tailored for imbalanced data. ECS mitigates these issues through adaptive pruning that preserves minority examples, class-sensitive partitioning aligned with skewed class distributions, and stratified graph-cut selection for diverse sampling. Experiments across multiple imbalanced datasets show that ECS improves generalization and substantially boosts minority-class accuracy compared to standard coreset methods.

The Combined Algorithm Selection and Hyperparameter Optimization (CASH) problem is pivotal in Automatic Machine Learning (AutoML). Most leading approaches combine Bayesian optimization with post-hoc ensemble building to create advanced AutoML systems. Bayesian optimization (BO) typically focuses on identifying a singular algorithm and its hyperparameters that outperform all other configurations. Recent developments have highlighted an oversight in prior CASH methods: the lack of consideration for diversity among the base learners of the ensemble. This oversight was overcome by explicitly injecting the search for diversity into the traditional CASH problem. However, despite recent developments, BO's limitation lies in its inability to directly optimize ensemble generalization error, offering no theoretical assurance that increased diversity correlates with enhanced ensemble performance. Our research addresses this gap by establishing a theoretical foundation that integrates diversity into the core of BO for direct ensemble learning. We explore a theoretically sound framework that describes the relationship between pair-wise diversity and ensemble performance, which allows our Bayesian optimization framework Optimal Diversity Bayesian Optimization (OptDivBO) to directly and efficiently minimize ensemble generalization error. OptDivBO guarantees an optimal balance between pairwise diversity and individual model performance, setting a new precedent in ensemble learning within CASH. Empirical results on 20 public datasets show that OptDivBO achieves the best average test ranks of 1.57 and 1.4 in classification and regression tasks.

Self and Semi-Supervised Learning have shown promising results in language and computer vision but are still underexplored in the context of tabular data. This paper focuses on exploring self and semi-supervised methods for tabular data. Towards this, we have proposed Auto-Tab Transformer, a method for training hierarchical transformers in a self and semi-supervised setup using redundancy reduction. The technique focuses on key aspects of self and semi-supervised learning: feature encoding, pre-training objective, training methodology and neural architecture. Performing extensive experiments on four publically accessible datasets, we show that Auto-Tab Transformer achieves state of the art (SOTA) results in the less labelled data domain. We conduct extensive ablation studies detailing the importance of all the components used.

Temporal Point Processes (TPP) are probabilistic generative frameworks. They model discrete event sequences localized in continuous time. Generally, real-life events reveal descriptive information, known as marks. Marked TPPs model time and marks of the event together for practical relevance. Conditioned on past events, marked TPPs aim to learn the joint distribution of the time and the mark of the next event. For simplicity, conditionally independent TPP models assume time and marks are independent given event history. They factorize the conditional joint distribution of time and mark into the product of individual conditional distributions. This structural limitation in the design of TPP models hurt the predictive performance on entangled time and mark interactions. In this work, we model the conditional inter-dependence of time and mark to overcome the limitations of conditionally independent models. We construct a multivariate TPP conditioning the time distribution on the current event mark in addition to past events. Besides the conventional intensity-based models for conditional joint distribution, we also draw on flexible intensity-free TPP models from the literature. The proposed TPP models outperform conditionally independent and dependent models in standard prediction tasks. Our experimentation on various datasets with multiple evaluation metrics highlights the merit of the proposed approach.

Predicting customer attrition (churn) is a well known problem in industries that provide services, like financial institutions, telecommunications, e-commerce, and retail. There are two kinds of attrition - active and passive (silent). Active attrition is usually associated with subscription-based business models, commonly seen in telecommunications and internet industries like Netflix. In industries like finance, retail, and ecommerce, we see the other kind of attrition - silent attrition where customers stop doing business without formal notice. This makes the silent attrition prediction problem even more challenging because it is difficult to differentiate between attrited and inactive customers. We focus our work on predicting silent attrition which is still under-explored in the payment card industry (i.e. Mastercard, Visa). The contribution of our work is threefold. First, we present a data-driven approach to define silent attrition as customer inactivity. Second, we discussed multiple procedures to generate synthetic data thereby preserving customers’ privacy. At last, we presented a comprehensive view of various machine learning (ML) pathways in which this churn prediction problem can be framed and solved; each requiring a specific feature engineering. We presented experimental results corresponding to each pathway to comparative analysis. We believe that this work to be beneficial to the researchers and ML practitioners who often have to deal with sensitive financial data but have limited permission to use it. In this direction, we demonstrated the use of synthetic data generation to reduce the risk of data leakage and other privacy concerns relating to ML models development.

The tremendous rise of cryptocurrency in the payment domain has unlocked huge opportunities but also raised numerous challenges in parallel involving cybercriminal activities like money laundering, terrorist financing, illegal and risky services, etc, owing to its anonymous and decentralized setup. The demand for building a more transparent cryptocurrency network, resilient to such activities, has risen extensively as more financial institutions look to incorporate it into their network. While a plethora of traditional machine learning and graph based deep learning techniques have been developed to detect illicit activities in a cryptocurrency transaction network, the challenge of generalization and robust model performance on future timesteps still exists. In this paper, we show that the model learned on transactional feature set provided in dataset (Elliptic Dataset) carry a temporal bias, i.e. they are highly dependent on the timesteps they occur. Deploying temporally biased models limits their performance on future timesteps. To address this, we propose a temporal debiasing technique using GNN based architecture that ensures generalization by adversarially learning between fraud1classification and temporal classification. The adversarial loss constructed optimizes the embeddings to ensure they 1.) perform well on fraud classification task 2.) does not contain temporal bias. The proposed architecture capture the underlying fraud patterns that remain consistent over time. We evaluate the performance of our proposed architecture on the Elliptic dataset and compare the performance with existing machine learning and graph-based architectures.1Fraud and illicit are used interchangeably in this paper

For decision-makers in the forecasting sector, decision processes like planning of facilities, an optimal day-to-day operation within the domain etc., are complex with several different levels to be considered. These decisions address widely different time horizons and aspects of the system, making it difficult to model. The advent of deep learning in forecasting solved the need for expensive hand-crafted features and deep domain knowledge. The work aims at giving a structure to the existing literature for time-series forecasting in deep learning. Based on the underlying structures of the technique, such as RNN, CNN, and Transformer, we have categorized various deep learning-based time series forecasting techniques and provided a consolidated report. Additionally, we have performed experiments to compare these techniques on 4 different publicly available datasets. Finally, based on these experiments, we provide an intuitive reasoning behind these performances. We believe that this work shall help the researchers in choosing relevant techniques for future research.

The pandemics like Coronavirus disease 2019 (COVID-19) require Governments and health professionals to make time-sensitive, critical decisions about travel restrictions and resource allocations. This paper identifies various factors that affect the spread of the disease using transaction data and proposes a model to predict the degree of spread of the disease and thus the number of medical resources required in upcoming weeks. We perform a region-wise analysis of these factors to identify the control measures that affect the minimal set of population. Our model also helps in estimating the surges in clinical demand and identifying when the medical resources would be saturated. Using this estimate, we suggest the preventive as well as corrective measures to avoid critical situations.

Interactive Voice Response (IVR) platforms have been widely deployed in resource-limited settings. These systems tend to afford asynchronous push interactions, and within the context of health, provide medication reminders, descriptions of symptoms and tips on self-management. Here, we present the development of an IVR system for resource-limited settings that enables real-time, synchronous interaction. Inspired by community radio, and calls for health systems that are truly local, we developed "Sehat ki Vaani". Sehat ki Vaani is a real-time IVR platform that enables hosting and participation in radio chat shows on community-led topics. We deployed Sehat ki Vaani with two communities in North India on topics related to the management of Type 2 diabetes and maternal health. Our deployments highlight the potential for synchronous IVR systems to offer community connection and localised sharing of experience, while also highlighting the complexity of producing, hosting and participating in radio shows in real time through IVR. We discuss the relative strengths and weaknesses of synchronous IVR systems, and highlight lessons learnt for interaction design in this area.

Waiting for the completion of a system process is an everyday experience. While waiting, system provides feedback to the user about ongoing process through temporal metaphors (Progress bar, Busy icons, etc.). One of the key performance requirement for temporal metaphors is to retain the user till the process completes. Researchers have evaluated these metaphors through subjective means, and objective assessment has not been well explored. In this paper, we present survival analysis as objective assessment method to evaluate temporal metaphors. Through a field experiment, we demonstrate the application of survival analysis and empirically establish that auditory progress bar (temporal metaphor for audio interfaces) works for callers of a distress helpline. To the best of our knowledge, it is the first study on distress callers. The paper further discusses the applicability of survival analysis for evaluating temporal metaphors and wait time experiments for other applications, tasks, and settings.

Abundance of mobile phones in developing nations makes them an effective tool to spread information and communication technology. Currently different mobile based platforms are being used in a variety of ICT contexts in developing nations. These systems have been built with different features according to the context in which they are used. After studying existing tools, we have developed a generic mobile-based ICT tool, named MVoice, which can serve in multiple contexts with minimal configuration changes. Our tool can also be extended easily to satisfy new requirements. In this paper, we present our experience and results from four real-world deployments, each in a different context, of MVoice.

Activist groups have taken up information dissemination and feedback collection as a means of rights advocacy in India. However, it is not easy given the difficulty in procuring and disseminating information at a large scale. Beneficiaries are often not able to help themselves as information systems are administration facing, because of poor literacy and the inability to access the Internet. Further, beneficiaries are not well informed of their rights and entitlements under different government schemes to know how and when to file grievances. We aim to solve these problems by designing and testing prototypes for information dissemination and feedback collection in various contexts. In our current prototype we describe an automated tool that sifts through the data on an MIS and conveys personalised information to the beneficiaries through voice calls. This is a work in progress, and our first exercise on providing MIS-extracted information to people through phone calls led to 70% of the beneficiaries who noticed a discrepancy in the data to agree to file a grievance on their behalf. We are continuing to scale the work, make it more automated, and run qualitative interviews with all stakeholders to understand causality linkages with transparency led grievance filing, assisted by appropriate ICTs, to increase accountability.

Increasing use of telephone devices has made the Interactive Voice Response (IVR), a technology for accessing information over phone, popular among the commercial organizations. IVR systems are used for critical applications like flight reservation, tele-banking, etc. which requires to have well tested IVR systems. Manual testing of an IVR application requires dialing number, listening and responding to voice prompts through key-press or speech.