Aarhus, Central Denmark Region, Denmark
Akhil Arora is an Assistant Professor of Computer Science at Aarhus University, a fellow of the Copenhagen Center for Social Data Science (SODAS), and a formal collaborator of the Wikimedia Foundation. Before this, he graduated with a PhD in Computer Science from EPFL. Akhil’s research lies at the intersection of natural language processing, human-centered AI, graph machine learning, and causal inference, with an overarching goal of developing the next generation of information systems that assist humans in seeking knowledge. In days of yore, Akhil spent close to five years in the industry working with the research labs of Xerox and American Express as a Research Scientist. His work on influence maximization has been recognized as the 8th most influential paper of SIGMOD 2017 by Paper Digest and received the 2018 ACM SIGMOD Most Reproducible Paper Award. He is a recipient of the prestigious EDIC Doctoral Fellowship, an alumnus of the coveted Heidelberg Laureate Forum, and a DAAD AINet fellow on human-centered AI. Further details are available on his website.
Head of the CLAN for AI Research on Language and Networks, or CLAN for short. I am looking for motivated PhD students to join the CLAN!! For details, please visit our lab website https://www.cs.au.dk/~clan/
Developing ML methods to model and improve human knowledge seeking on the Web. - LLMNav, a framework for simulating human website browsing behavior. - First privacy-preserving generative model of human Web browsing behavior. - Natural experiment to measure the causal effect of de-orphanization on article visibility in more than 300 language versions of Wikipedia. - Eigenthemes, the state-of-the-art unsupervised entity linker with 700x faster inference and comparable efficacy to transformer-based alternatives. - PARIS+, a probabilistic model with superior efficacy, 1000x faster inference, and 10x smaller memory footprint than neural methods for entity alignment. - Publications in EMNLP, NAACL, SIGIR, WSDM, VLDB, and ICWSM.
Devised scalable deep learning algorithms for the credit card fraud and risk assessment business of American Express leading to multi-million-dollar impact. - BAE, an ensemble of data bagged Autoencoders for improving credit card fraud detection - NL2SQL, sequence-to-sequence model for translating natural language instructions to SQL queries for descriptive analytics. - TextRisk, an LSTM-based model trained on transcribed customer care conversations, which improved delinquency prediction performance by 5%.
Led a team of 10 research scientists and engineers for devising scalable data management algorithms and machine learning models to solve a gamut of complex real-world problems for the Customer-care and Health-care business of Xerox. Highlights: - GaBiD, a 360-degree customer journey analytics framework with novel models for churn prediction, root-cause identification, and prevention. - KEO, information extraction framework to construct enterprise knowledge graphs. - Multiple patent applications and publications in SIGMOD, VLDB, and WWW.