Xiaolin Xu

About

Xiaolin Xu is an Associate Professor in the Department of Electrical and Computer Engineering at Northeastern University College of Engineering. His research focuses on security, machine learning, and AI security/privacy, energy-efficient deep learning, computer architecture, FPGA, embedded systems, and VLSI. He has led multiple research projects funded by the National Science Foundation, including designing and optimizing tiny vector symbolic architectures for ultra-efficient inference on tiny devices, securing brain-inspired hyperdimensional computing against attacks for edge devices, and accelerating privacy-preserving machine learning as a service from algorithm to hardware. Professor Xu has been recognized for his contributions to the field, receiving awards such as the IEEE/ACM Design Automation Conference (DAC) Under-40 Innovators Award in 2025. His work involves developing secure and robust machine learning hardware accelerators, securing brain-inspired hyperdimensional computing, and enhancing the security of scientific cyberinfrastructures. He collaborates with various institutions and industry partners, contributing to advancing trusted hardware, AI chip security, and energy-efficient AI solutions.

Research topics

• Physics
• Particle physics
• Astrophysics
• Astronomy
• Nuclear physics
• Artificial Intelligence
• Computer Science
• Optics

Selected publications

• Evidence for neutrino emission from the nearby active galaxy NGC 1068

  Science · 2022 · 446 citations

  A supermassive black hole, obscured by cosmic dust, powers the nearby active galaxy NGC 1068. Neutrinos, which rarely interact with matter, could provide information on the galaxy's active core. We searched for neutrino emission from astrophysical objects using data recorded with the IceCube neutrino detector between 2011 and 2020. The positions of 110 known gamma-ray sources were individually searched for neutrino detections above atmospheric and cosmic backgrounds. We found that NGC 1068 has a…

  DOI

• Time-Integrated Neutrino Source Searches with 10 Years of IceCube Data

  Physical Review Letters · 2020 · 433 citations

  This Letter presents the results from pointlike neutrino source searches using ten years of IceCube data collected between April 6, 2008 and July 10, 2018. We evaluate the significance of an astrophysical signal from a pointlike source looking for an excess of clustered neutrino events with energies typically above ∼1 TeV among the background of atmospheric muons and neutrinos. We perform a full-sky scan, a search within a selected source catalog, a catalog population study, and three stacked Ga…

  DOI

• IceCube high-energy starting event sample: Description and flux characterization with 7.5 years of data

  Physical review. D/Physical review. D. · 2021 · 368 citations

  The IceCube Neutrino Observatory has established the existence of a high-energy all-sky neutrino flux of astrophysical origin. This discovery was made using events interacting within a fiducial region of the detector surrounded by an active veto and with reconstructed energy above 60 TeV, commonly known as the high-energy starting event sample (HESE). We revisit the analysis of the HESE sample with an additional 4.5 years of data, newer glacial ice models, and improved systematics treatment. Thi…

  DOI

• Graph Neural Networks for low-energy event classification & reconstruction in IceCube

  Journal of Instrumentation · 2022 · 38 citations

  Abstract IceCube, a cubic-kilometer array of optical sensors built to detect atmospheric and astrophysical neutrinos between 1 GeV and 1 PeV, is deployed 1.45 km to 2.45 km below the surface of the ice sheet at the South Pole. The classification and reconstruction of events from the in-ice detectors play a central role in the analysis of data from IceCube. Reconstructing and classifying events is a challenge due to the irregular detector geometry, inhomogeneous scattering and absorption of light…

  DOI

• Deep Graph Neural Point Process For Learning Temporal Interactive Networks

  ArXiv.org · 2025-08-17

  preprintOpen access

  Learning temporal interaction networks(TIN) is previously regarded as a coarse-grained multi-sequence prediction problem, ignoring the network topology structure influence. This paper addresses this limitation and a Deep Graph Neural Point Process(DGNPP) model for TIN is proposed. DGNPP consists of two key modules: the Node Aggregation Layer and the Self Attentive Layer. The Node Aggregation Layer captures topological structures to generate static representation for users and items, while the Se…

  PublisherOA PDF

Frequent coauthors

• S. R. Klein

  866 shared

• G. M. Spiczak

  701 shared

• G. C. Hill

  Providence College

  697 shared

• G. T. Przybylski

  Lawrence Berkeley National Laboratory

  696 shared

• D. R. Williams

  687 shared

• Juan Pablo Yáñez

  672 shared

• D. J. Koskinen

  665 shared

• G. S. Japaridze

  636 shared

Labs

• Xiaolin Xu LabPI

Awards & honors

• 2025 IEEE/ACM Design Automation Conference (DAC) Under-40 In…
• NSF CAREER Award: Securing Reconfigurable Hardware Accelerat…
• NSF Award for Securing Scientific Cyberinfrastructures From…
• $560K NSF grant for Designing and Optimizing Tiny Vector Sym…
• $600,000 NSF grant for Securing Brain-Inspired Hyperdimensio…

Similar researchers at Northeastern University

• Ravinder Dahiyah-81
• Milica Stojanovich-63
• Engin Kirdah-62