BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Los_Angeles
X-LIC-LOCATION:America/Los_Angeles
BEGIN:DAYLIGHT
TZOFFSETFROM:-0800
TZOFFSETTO:-0700
TZNAME:PDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0700
TZOFFSETTO:-0800
TZNAME:PST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20240626T180034Z
LOCATION:3001\, 3rd Floor
DTSTART;TZID=America/Los_Angeles:20240626T111500
DTEND;TZID=America/Los_Angeles:20240626T113000
UID:dac_DAC 2024_sess102_RESEARCH390@linklings.com
SUMMARY:Leanor: A Learning-Based Accelerator for Efficient Approximate Nea
 rest Neighbor Search via Reduced Memory Access
DESCRIPTION:Research Manuscript\n\nYi Wang, Huan Liu, Jianan Yuan, Jiaxian
  Chen, Tianyu Wang, Chenlin Ma, and Rui Mao (Shenzhen University)\n\nAppro
 ximate Nearest Neighbor Search (ANNS) is a classical problem in data scien
 ce. ANNS is both computationally-intensive and memory-intensive. As a typi
 cal implementation of ANNS, Inverted File with Product Quantization (IVFPQ
 ) has the properties of high precision and rapid processing. However, the 
 traversal of non-nearest neighbor vectors in IVFPQ leads to redundant memo
 ry accesses. This significantly impacts retrieval efficiency. A promising 
 approach involves the utilization of learned indexes, leveraging insights 
 from data distribution to optimize search efficiency. Existing learned ind
 exes are primarily customized for low-dimensional data. How to tackle ANNS
  in high-dimensional vectors is a challenging issue. \n\nThis paper introd
 uces Leanor, a learned index-based accelerator for the filtering of non-ne
 arest neighbor vectors within the IVFPQ framework. Leanor minimizes redund
 ant memory accesses, thereby enhancing retrieval efficiency. Leanor incorp
 orates a dimension reduction component, mapping vectors to one-dimensional
  keys and organizing them in a specific order. Subsequently, the learned i
 ndex leverages this ordered representation for rapid predictions. To enhan
 ce result accuracy, we conduct a thorough analysis of model errors and int
 roduce a specialized index structure named learned index forest (LIF). The
  experimental results show that, compared to representative approaches, Le
 anor can effectively filter out non-neighboring vectors within IVFPQ, lead
 ing to a substantial enhancement in retrieval efficiency.\n\nTopic: AI\n\n
 Keyword: AI/ML Algorithms\n\nSession Chairs: Parivesh Choudhary (Synopsys)
  and Anca Molnos (CEA-List)
END:VEVENT
END:VCALENDAR