LIBOL: A Library for Online Learning Algorithms
Authors: Steven C.H. Hoi, Jialei Wang, Peilin Zhao
JMLR 2014 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | To illustrate the online learning procedure, we take two data sets from the LIBSVM website, including one small data set svmguide3 with 1243 instances and one large data set ijcnn1 with 141,691 instances. In the following example, we use the default Perceptron algorithm to demo the usage of LIBOL for a binary classification ( bc ) task: $ demo( bc , Perceptron , svmguide3 ) The results output by the above command are summarized as follows: Algorithm: mistake rate nb of updates cpu time (seconds) Perceptron 0.3318 +/0.0118 412.45 +/14.66 0.0516 +/0.0008. |
| Researcher Affiliation | Academia | Steven C.H. Hoi EMAIL Jialei Wang EMAIL Peilin Zhao EMAIL School of Computer Engineering Nanyang Technological University Singapore 639798 |
| Pseudocode | Yes | Algorithm 1: LIBOL: Online Learning Framework for Linear Classification. 1 Initialize: w1 = 0 2 for t = 1, 2, . . . , T do 3 The learner receives an incoming instance: xt X; 4 The learner predicts the class label: ˆyt = sgn(f(xt; wt)); 5 The true class label is revealed from the environment: yt Y; 6 The learner calculates the suffered loss: ℓ(wt; (xt, yt)); 7 if ℓ(wt; (xt, yt)) > 0 then 8 The learner updates the classification model: 9 wt+1 wt + (wt; (xt, yt)); |
| Open Source Code | Yes | The LIBOL software is available at http://libol.stevenhoi.org/. |
| Open Datasets | Yes | To illustrate the online learning procedure, we take two data sets from the LIBSVM website, including one small data set svmguide3 with 1243 instances and one large data set ijcnn1 with 141,691 instances. |
| Dataset Splits | No | The paper uses the svmguide3 and ijcnn1 datasets from the LIBSVM website but does not provide specific details on how these datasets were split into training, validation, or test sets, or if cross-validation was used. It describes an online learning process where data is processed sequentially, but not explicit splits for reproduction. |
| Hardware Specification | No | The paper mentions 'cpu time (seconds)' in its experimental results, but it does not provide any specific details about the CPU model, GPU, memory, or any other hardware components used for running the experiments. |
| Software Dependencies | No | The paper mentions 'MATLAB library' and 'C/C++ implementation for the core functions' and that 'The data formats used by this software are compatible with popular machine learning and data mining packages, such as LIBSVM, SVM-light, and WEKA, etc.'. However, it does not provide specific version numbers for MATLAB, C/C++ compilers, or any of the mentioned compatible packages. |
| Experiment Setup | No | The paper states that the `run_experiment` command 'will run side-by-side comparison of varied online learning algorithms on the given data set fully automatically, including all the parameter settings and selection.' However, it does not explicitly list these 'parameter settings and selection' (e.g., learning rates, epochs, specific algorithm hyperparameters) within the text. |