We investigate polarity reversals in the geodynamo using a rotating, convection-driven dynamo model. As flow rapidly rotating convection is dominated by columns aligned with axis of rotation, focus on dynamics columnar vortices. By studying growth seed magnetic field to stable axial dipole field, we show that acts ways significantly enhance relative helicity between cyclonic and anticyclonic This asymmetry hallmark dipolar dynamo. Strong buoyancy, other hand, offsets effect establishing...

Entity Alignment (EA) is the task of recognizing same entity present in different knowledge bases. Recently, embedding-based EA techniques have established dominance where alignment done based on closeness latent space. Graph Neural Networks (GNN) gained popularity as embedding module due to its ability learn entities' representation their local sub-graph structures. Although GNN shows promising results, limited works aimed capture relations while considering global importance and relative...

Credit card fraud detection is arguably the most critical use case of machine learning for any payment system. Deep neural networks and tree-based classifiers can provide state-of-the-art performance classification. However, we try to emphasize that these models have serious vulnerabilities need be addressed. Studies show it possible fool with curated input samples known as adversarial examples. Attackers examples deceive deployed by institutions, causing considerable financial harm. We feel...

Demand forecasting is a fundamental aspect of inventory and supply chain management. Due to the sporadic nature demand, demand involves dealing with intermittent time series in domains such as retail, manufacturing. Conventional methods do not work well for due inherent sparsity series. Researchers have proposed multiple deal Croston its variants. Our aims provide an insight into various traditionally known We also explored deep learning that been recent literature. These are thoroughly...

The large variety of digital payment choices available to consumers today has been a key driver e-commerce transactions in the past decade. Unfortunately, this also given rise cybercriminals and fraudsters who are constantly looking for vulnerabilities these systems by deploying increasingly sophisticated fraud attacks. A typical detection system employs standard supervised learning methods where focus is on maximizing recall rate. However, we argue that such formulation can lead sub-optimal...

Entity Alignment (EA) identifies entities across databases that refer to the same entity. Knowledge graph-based embedding methods have recently dominated EA techniques. Such map a low-dimension space and align them based on their similarities. With corpus of methodologies growing rapidly, this paper presents comprehensive analysis various existing methods, elaborating applications limitations. Further, we distinguish underlying algorithms information they incorporate learn entity...

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

Large Transformer based models have provided state-of-the-art results on a variety of Natural Language Processing (NLP) tasks. While these perform exceptionally well wide range NLP tasks, their usage in Sequence Labeling has been mostly muted. Although pretrained such as BERT and XLNet successfully employed input representation, the use model context encoder for sequence labeling is still minimal, most recent works recurrent architecture encoder. In this paper, we compare performance...

The inception of modeling contextual information using models such as BERT, ELMo, and Flair has significantly improved representation learning for words. It also given SOTA results in almost every NLP task - Machine Translation, Text Summarization Named Entity Recognition, to name a few. In this work, addition these dominant context-aware representations, we propose Knowledge Aware Representation Learning (KARL) Network Recognition (NER). We discuss the challenges existing methods...

The adversarial attack is a pivotal field of research. These attacks can help the organization identify model vulnerabilities and save them from financial reputation loss. Researchers have already explored several on computer vision applications. However, these fail when it comes to transactional dataset-which has added constraints limited attempts scalability for large feature space containing both categorical merchant code continuous transaction amount. existing literature tabular domain...

Global losses due to payment fraud have tripled from $9.84 billion in 2011 $32.39 2020 and are expected reach $40.62 by 2027. In addition the financial losses, negatively impacts brand reputation leads a bad customer experience. Advanced machine learning has been actively adopted tackle detection problem at scale. However, scarcity of open datasets less reproducible research, especially payments domain. We released synthetic transactional dataset, FraudAmmo, containing 3 million...

Planning based on long and short term time series forecasts is a common practice across many industries. In this context, temporal aggregation reconciliation techniques have been useful in improving forecasts, reducing model uncertainty, providing coherent forecast different horizons. However, an underlying assumption spanning all these the complete availability of data levels hierarchy, while offers mathematical convenience but most low frequency partially completed it not available...

In the context of time series forecasting, it is a common practice to evaluate multiple methods and choose one these or an ensemble for producing best forecasts. However, choosing among different ensembles over remains challenging task that undergoes combinatorial explosion as number increases. demand forecasting revenue this challenge further exacerbated by large well limited historical data points available due changing business context. Although deep learning aim simultaneously forecast...

Recent years of research in Natural Language Processing (NLP) have witnessed dramatic growth training large models for generating context-aware language representations. In this regard, numerous NLP systems leveraged the power neural network-based architectures to incorporate sense information embeddings, resulting Contextualized Word Embeddings (CWEs). Despite progress, community has not any significant work performing a comparative study on contextualization such architectures. This paper...