From 07b7e8f7f22d45837f2261067ff510871482bfa7 Mon Sep 17 00:00:00 2001 From: Ommac Date: Fri, 3 Mar 2017 18:06:25 +0100 Subject: [PATCH] updated description and maxiter to match the paper's experimental setting --- README.md | 20 ++++++++++++++++++++ main.m | 4 ++-- 2 files changed, 22 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index 26e78ee..5a61791 100644 --- a/README.md +++ b/README.md @@ -1,2 +1,22 @@ # incremental_multiclass_RLSC Incremental regularized least squares for multiclass classification with recoding, extension to new classes and fixed update complexity. + +Code for the experiments of the paper on the MNIST handwritten digits dataset: + "Incremental Robot Learning of New Objects with Fixed Update Time" + Raffaello Camoriano, Giulia Pasquale, Carlo Ciliberto, Lorenzo Natale, Lorenzo Rosasco, Giorgio Metta + ICRA 2017 + +Abstract + We consider object recognition in the context of lifelong learning, where a robotic agent learns to discriminate between a growing number of object classes as it accumulates experience about the environment. We propose an incremental variant of the Regularized Least Squares for Classification (RLSC) algorithm, and exploit its structure to seamlessly add new classes to the learned model. The presented algorithm addresses the problem of having an unbalanced proportion of training examples per class, which occurs when new objects are presented to the system for the first time. + We evaluate our algorithm on both a machine learning benchmark dataset and two challenging object recognition tasks in a robotic setting. Empirical evidence shows that our approach achieves comparable or higher classification performance than its batch counterpart when classes are unbalanced, while being significantly faster. + +Copyright (c) 2017 +Istituto Italiano di Tecnologia, Genoa, Italy +R. Camoriano, G. Pasquale, C. Ciliberto, L. Natale, L. Rosasco, G. Metta +All rights reserved. + +Clone and run main.m + +To modify the experimental setting, edit "dataConf_MNIST_inc.m" and/or the "Experimental setup" section of "main.m" + +Tested on MATLAB R2014b \ No newline at end of file diff --git a/main.m b/main.m index 405b47b..919240d 100644 --- a/main.m +++ b/main.m @@ -46,11 +46,11 @@ close all; confIncremental; -%% Experiments setup +%% Experimental setup run_inc_rlsc_yesrec = 1; % Incremental RLSC with recoding computeTestAcc = 1; % Flag for test accuracy computation trainPart = 0.8; % Training set part (1 - validation_percentage) -maxiter = 100; % Maximum number of updates +maxiter = 300; % Maximum number of updates numrep = 2; % Number of repetitions of the experiment saveResult = 1; % Save final results flag