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:20240626T180002Z
LOCATION:3001\, 3rd Floor
DTSTART;TZID=America/Los_Angeles:20240626T103000
DTEND;TZID=America/Los_Angeles:20240626T120000
UID:dac_DAC 2024_sess102@linklings.com
SUMMARY:AI Paradigms beyond Deep Neural Networks
DESCRIPTION:Research Manuscript\n\nAs hardware performance slows in the po
 st-Moore era, we need to explore new computing paradigms to improve the en
 ergy efficiency of AI. This session delves into cutting-edge algorithm res
 earch beyond DNNs and its practical applications. The presentations in thi
 s session discuss Optical Neural Networks, Random Forests, Hyper-Dimension
 al Computing, Approximate Nearest Neighbor Search, and Graph Neural Networ
 ks.\n\nSMORE: Similarity-Based Hyperdimensional Domain Adaptation for Mult
 i-Sensor Time Series Classification\n\nMany real-world applications of the
  Internet of Things (IoT) employ machine learning (ML) algorithms to analy
 ze time series information collected by interconnected sensors. However, d
 istribution shift, a fundamental challenge in data-driven ML, arises when 
 a model is deployed on a data distribution ...\n\n\nJunyao Wang and Mohamm
 ad Al Faruque (University of California, Irvine)\n---------------------\nG
 NNavigator: Towards Adaptive Training of Graph Neural Networks via Automat
 ic Guideline Exploration\n\nGraph Neural Networks (GNNs) succeed significa
 ntly in many applications recently. However, balancing GNNs training runti
 me cost, memory consumption, and attainable accuracy for various applicati
 ons is non-trivial. Previous training methodologies suffer from inferior a
 daptability and lack a unified t...\n\n\nTong Qiao, Jianlei Yang, Yingjie 
 Qi, and Ao Zhou (Beihang University); Chen Bai and Bei Yu (The Chinese Uni
 versity of Hong Kong); and Weisheng Zhao and Chunming Hu (Beihang Universi
 ty)\n---------------------\nZeroth-Order Optimization of Optical Neural Ne
 tworks with Linear Combination Natural Gradient and Calibrated Model\n\nOp
 tical neural networks (ONNs) have attracted great attention due to their l
 ow energy consumption and high-speed processing. The usual neural network 
 training scheme leads to poor performance for ONNs because of their specia
 l parameterization and fabrication variations. This paper contributes to e
 xt...\n\n\nHiroshi Sawada, Kazuo Aoyama, and Kohei Ikeda (NTT Corporation)
 \n---------------------\nLess is More: Hop-Wise Graph Attention for Scalab
 le and Generalizable Learning on Circuits\n\nWhile graph neural networks (
 GNNs) have gained popularity for learning circuit representations in vario
 us electronic design automation (EDA) tasks, they face challenges in scala
 bility when applied to large graphs and exhibit limited generalizability t
 o new designs. These limitations make them less p...\n\n\nChenhui Deng and
  Zichao Yue (Cornell University); Cunxi Yu (University of Maryland, Colleg
 e Park); Gokce Sarar, Ryan Carey, and Rajeev Jain (Qualcomm); and Zhiru Zh
 ang (Cornell University)\n---------------------\nLeanor: A Learning-Based 
 Accelerator for Efficient Approximate Nearest Neighbor Search via Reduced 
 Memory Access\n\nApproximate Nearest Neighbor Search (ANNS) is a classical
  problem in data science. ANNS is both computationally-intensive and memor
 y-intensive. As a typical implementation of ANNS, Inverted File with Produ
 ct Quantization (IVFPQ) has the properties of high precision and rapid pro
 cessing. However, the...\n\n\nYi Wang, Huan Liu, Jianan Yuan, Jiaxian Chen
 , Tianyu Wang, Chenlin Ma, and Rui Mao (Shenzhen University)\n------------
 ---------\nOrder-Preserving Cryptography for the Confidential Inference in
  Random Forests: FPGA Design and Implementation\n\nPrior work has addresse
 d the problem of confidential inference in decision trees. Both traditiona
 l order-preserving cryptography and order-preserving NTRU cryptography hav
 e been used to ensure data and model privacy in decision trees. Furthermor
 e, FPGA architectures and implementations have been pro...\n\n\nRupesh Kar
 n (New York University), Kashif Nawaz (Technology Innovation Institute), a
 nd Ibrahim (Abe) Elfadel (Khalifa University)\n\nTopic: AI\n\nKeyword: AI/
 ML Algorithms\n\nSession Chairs: Parivesh Choudhary (Synopsys) and Anca Mo
 lnos (CEA-List)
END:VEVENT
END:VCALENDAR