Showing 1–2 of 2 results for author: Furusawa, T

Mean Field Theory in Deep Metric Learning

Authors: Takuya Furusawa

Abstract: In this paper, we explore the application of mean field theory, a technique from statistical physics, to deep metric learning and address the high training complexity commonly associated with conventional metric learning loss functions. By adapting mean field theory for deep metric learning, we develop an approach to design classification-based loss functions from pair-based ones, which can be con… ▽ More In this paper, we explore the application of mean field theory, a technique from statistical physics, to deep metric learning and address the high training complexity commonly associated with conventional metric learning loss functions. By adapting mean field theory for deep metric learning, we develop an approach to design classification-based loss functions from pair-based ones, which can be considered complementary to the proxy-based approach. Applying the mean field theory to two pair-based loss functions, we derive two new loss functions, MeanFieldContrastive and MeanFieldClassWiseMultiSimilarity losses, with reduced training complexity. We extensively evaluate these derived loss functions on three image-retrieval datasets and demonstrate that our loss functions outperform baseline methods in two out of the three datasets. △ Less

Submitted 27 June, 2023; originally announced June 2023.

Comments: 15 pages

Area-Efficient Selective Multi-Threshold CMOS Design Methodology for Standby Leakage Power Reduction

Authors: Takeshi Kitahara, Naoyuki Kawabe, Fimihiro Minami, Katsuhiro Seta, Toshiyuki Furusawa

Abstract: This paper presents a design flow for an improved selective multi-threshold(Selective-MT) circuit. The Selective-MT circuit is improved so that plural MT-cells can share one switch transistor. We propose the design methodology from RTL(Register Transfer Level) to final layout with optimizing switch transistor structure. This paper presents a design flow for an improved selective multi-threshold(Selective-MT) circuit. The Selective-MT circuit is improved so that plural MT-cells can share one switch transistor. We propose the design methodology from RTL(Register Transfer Level) to final layout with optimizing switch transistor structure. △ Less

Submitted 25 October, 2007; originally announced October 2007.

Comments: Submitted on behalf of EDAA (http://www.edaa.com/)

Journal ref: Dans Design, Automation and Test in Europe - DATE'05, Munich : Allemagne (2005)