skip to main content

Explore Research Products in the PAR
It may take a few hours for recently added research products to appear in PAR search results.


Title: SNNOpt: An Application-Specific Design Framework for Spiking Neural Networks
We propose a systematic application-specific hardware design methodology for designing Spiking Neural Network (SNN), SNNOpt, which consists of three novel phases: 1) an Olliver-Ricci-Curvature (ORC)-based architecture-aware network partitioning, 2) a reinforcement learning mapping strategy, and 3) a Bayesian optimization algorithm for NoC design space exploration. Experimental results show that SNNOpt achieves a 47.45% less runtime and 58.64% energy savings over state-of-the-art approaches.  more » « less
Award ID(s):
1932620
PAR ID:
10481611
Author(s) / Creator(s):
; ; ; ; ; ; ; ;
Publisher / Repository:
IEEE
Date Published:
Journal Name:
2023 IEEE 5th International Conference on Artificial Intelligence Circuits and Systems (AICAS)
ISBN:
979-8-3503-3267-4
Page Range / eLocation ID:
1 to 5
Format(s):
Medium: X
Location:
Hangzhou, China
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ( , Schloss Dagstuhl – Leibniz-Zentrum für Informatik)
    Chan, Timothy ; Fischer, Johannes ; Iacono, John ; Herman, Grzegorz (Ed.)
    We consider two-cost network design models in which edges of the input graph have an associated cost and length. We build upon recent advances in hop-constrained oblivious routing to obtain two sets of results. We address multicommodity buy-at-bulk network design in the nonuniform setting. Existing poly-logarithmic approximations are based on the junction tree approach [Chekuri et al., 2010; Guy Kortsarz and Zeev Nutov, 2011]. We obtain a new polylogarithmic approximation via a natural LP relaxation. This establishes an upper bound on its integrality gap and affirmatively answers an open question raised in [Chekuri et al., 2010]. The rounding is based on recent results in hop-constrained oblivious routing [Ghaffari et al., 2021], and this technique yields a polylogarithmic approximation in more general settings such as set connectivity. Our algorithm for buy-at-bulk network design is based on an LP-based reduction to h-hop constrained network design for which we obtain LP-based bicriteria approximation algorithms. We also consider a fault-tolerant version of h-hop constrained network design where one wants to design a low-cost network to guarantee short paths between a given set of source-sink pairs even when k-1 edges can fail. This model has been considered in network design [Luis Gouveia and Markus Leitner, 2017; Gouveia et al., 2018; Arslan et al., 2020] but no approximation algorithms were known. We obtain polylogarithmic bicriteria approximation algorithms for the single-source setting for any fixed k. We build upon the single-source algorithm and the junction-tree approach to obtain an approximation algorithm for the multicommodity setting when at most one edge can fail. 
    more » « less
  2. ; ; ( , Proceedings of the 2020 Conference on Robot Learning)
    The 3D shape of a robot’s end-effector plays a critical role in determining it’s functionality and overall performance. Many of today’s industrial applications rely on highly customized gripper design for a given task to ensure the system’s robustness and accuracy. However, the process of manual hardware design is both costly and time-consuming, and the quality of the design is also dependent on the engineer’s experience and domain expertise, which can easily be out-dated or inaccurate. The goal of this paper is to use machine learning algorithms to automate this design process and generate task-specific gripper designs that satisfy a set of pre-defined design objectives. We model the design objectives by training a Fitness network to predict their values for a pair of gripper fingers and a grasp object. This Fitness network is then used to provide training supervision to a 3D Generative network that produces a pair of 3D finger geometries for the target grasp object. Our experiments demonstrate that the proposed 3D generative design framework generates parallel jaw gripper finger shapes that achieve more stable and robust grasps as compared to other general-purpose and task-specific gripper design algorithms. 
    more » « less
  3. ; ; ; ; ; ; ; ( , Giant)
    Current polymer network design suffers from intrinsic trade-offs, where polymer networks with high modulus often turn out to be in short of stretchability or fracture toughness. Here, we show a novel polymer network design through polymer-nanoparticle alternating hybrids that enable integrating the non-polymeric colloid deformation into polymer network design. The new class of polymer network exhibits colloidal yielding at small deformation before conformational change at higher elongation ratios, enabling simultaneous achievement of high Young’s modulus of E≈10-50 MPa, high yield strength of σ_Y~ 3-5 MPa, large stretchability of λ~7-10, and high fracture energy density of Γ~30 MJ/m^3. These results demonstrate a successful strategy to decouple the molecular mechanics for yield from that for stretchability or toughness, leading to new polymer networks design. 
    more » « less
  4. ; ; ; ; ; ; ( , Journal of Educational Computing Research)

    Integrating engineering design into K-12 curricula is increasingly important as engineering has been incorporated into many STEM education standards. However, the ill-structured and open-ended nature of engineering design makes it difficult for an instructor to keep track of the design processes of all students simultaneously and provide personalized feedback on a timely basis. This study proposes a Bayesian network model to dynamically and automatically assess students’ engagement with engineering design tasks and to support formative feedback. Specifically, we applied a Bayesian network to 111 ninth-grade students’ process data logged by a computer-aided design software program that students used to solve an engineering design challenge. Evidence was extracted from the log files and fed into the Bayesian network to perform inferential reasoning and provide a barometer of their performance in the form of posterior probabilities. Results showed that the Bayesian network model was competent at predicting a student’s task performance. It performed well in both identifying students of a particular group (recall) and ensuring identified students were correctly labeled (precision). This study also suggests that Bayesian networks can be used to pinpoint a student’s strengths and weaknesses for applying relevant science knowledge to engineering design tasks. Future work of implementing this tool within the computer-aided design software will provide instructors a powerful tool to facilitate engineering design through automatically generating personalized feedback to students in real time.

     
    more » « less
  5. ( , Communications in computer and information science)
    Wasson, B. (Ed.)
    Visualization plays an important role in Epistemic Network Analysis (ENA), not only in graphical representation but also to facilitate interpretation and communicate research findings. However, there is no published description of the design features behind ENA network graphs. This paper provides this description from a graphic design perspective, focusing on the design principles that make ENA network graphs aesthetically pleasing and intuitive to understand. By reviewing graphic design principles and examining other extant network visualizations, we show how the current ENA network graphs highlight the most important network characteristics and facilitate sense-making. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email