BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:In-Network Machine Learning for Market Prediction Using Limit Orde
 r Books - Xinpeng Hong\, University of Oxford
DTSTART:20230518T140000Z
DTEND:20230518T150000Z
UID:TALK201310@talks.cam.ac.uk
CONTACT:Ryan Gibb
DESCRIPTION:Abstract: Machine learning (ML) is driving the evolution of al
 gorithmic trading\, but conflicts with the demand for fast execution speed
 . Although both aim to drive higher profitability\, embedding more powerfu
 l ML approaches and lowering trading latencies are hard to achieve simulta
 neously. Offloading ML inference to programmable network devices\, also ca
 lled in-network ML\, provides a delicate balance between the two ends of t
 his trade-off. In this talk\, I present LOBIN\, providing ML-based market 
 movement prediction using high-frequency market data feeds. LOBIN builds l
 imit order books and conducts ML-based inference within programmable switc
 hes. This talk describes our solution to the challenging task of mapping L
 OB constructs and suitable ML models to a commodity switch. Compared with 
 existing solutions\, LOBIN predicts stock price movements with lower laten
 cy\, higher throughput\, and a minor impact on ML performance.        \n\n
 Bio: Xinpeng Hong is a second-year DPhil student in the Computing Infrastr
 ucture Group at the University of Oxford. His research interest lies in ti
 me-sensitive applications of in-network ML.  
LOCATION:FW26 and https://cl-cam-ac-uk.zoom.us/j/97216272378?pwd=M2diTFhMT
 nppckJtNWhFVTBKK0REZz09
END:VEVENT
END:VCALENDAR