This paper presents a method to more efficiently set the margin parameters of margin-based softmax for training a visual place recognition model. We observe an inconsistency in sample emphasis when using the conventional margin-based softmax loss and propose to solve this by utilizing two gradient scaling curves. We propose to assign more emphasis on the samples that currently come as inputs by analyzing the gradient scaling curve in terms of the angle between two vectors: the feature vector and its ground truth class vector. The gradient scaling curve shows that the model mainly emphasizes the semi-trained samples while it gives relatively small emphasis on the well-trained samples. We propose a method to adjust both angular and additive margin parameters to control the shape of the gradient scaling curve and highlight the samples that are currently being trained on. Moreover, gradient scaling is very sensitive to the type of optimizer and the learning rate. In this paper, we examine our gradient scaling effect in various optimization environments and find out the best configuration guideline for our novel loss.


    Zugriff

    Zugriff prüfen

    Verfügbarkeit in meiner Bibliothek prüfen

    Bestellung bei Subito €


    Exportieren, teilen und zitieren



    Titel :

    Adaptively Marginalizing Cosine Logits for Effectively Learning Deep Visual Place Representation in Loop Closure Detection


    Weitere Titelangaben:

    Lect. Notes in Networks, Syst.


    Beteiligte:
    Abdul Majeed, Anwar P.P. (Herausgeber:in) / Yap, Eng Hwa (Herausgeber:in) / Liu, Pengcheng (Herausgeber:in) / Huang, Xiaowei (Herausgeber:in) / Nguyen, Anh (Herausgeber:in) / Chen, Wei (Herausgeber:in) / Kim, Ue-Hwan (Herausgeber:in) / Kim, Jae-woo (Autor:in) / Kim, Ue-hwan (Autor:in)

    Kongress:

    International Conference on Robot Intelligence Technology and Applications ; 2023 ; Taicang December 06, 2023 - December 08, 2023



    Erscheinungsdatum :

    22.11.2024


    Format / Umfang :

    17 pages





    Medientyp :

    Aufsatz/Kapitel (Buch)


    Format :

    Elektronische Ressource


    Sprache :

    Englisch




    Logits DeConfusion with CLIP for Few-Shot Learning

    Li, Shuo / Liu, Fang / Hao, Zehua et al. | ArXiv | 2025

    Freier Zugriff

    Knowledge Distillation with Refined Logits

    Sun, Wujie / Chen, Defang / Lyu, Siwei et al. | ArXiv | 2024

    Freier Zugriff

    Exploiting Features and Logits in Heterogeneous Federated Learning

    Chan, Yun-Hin / Ngai, Edith C. -H. | ArXiv | 2022

    Freier Zugriff

    NormKD: Normalized Logits for Knowledge Distillation

    Chi, Zhihao / Zheng, Tu / Li, Hengjia et al. | ArXiv | 2023

    Freier Zugriff

    Principal components analysis of the logits of the survivorship function

    Mitra, S. / Levin, M. L. / International Union for the Scientific Study of Population | British Library Conference Proceedings | 1993