Gradient based sample selection for online continual learning

Authors: Rahaf Aljundi, Min Lin, Baptiste Goujaud, Yoshua Bengio

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental 4 Experiments This section serves to validate our approach and show its effectiveness at dealing with continual learning problems where task boundaries are not available.
Researcher Affiliation Academia Rahaf Aljundi KU Leuven rahaf.aljundi@gmail.com Min Lin Mila mavenlin@gmail.com Baptiste Goujaud Mila baptiste.goujaud@gmail.com Yoshua Bengio Mila yoshua.bengio@mila.quebec
Pseudocode Yes Algorithm 1 IQP Sample Selection; Algorithm 2 Greedy Sample Selection
Open Source Code Yes The code is available at https://github.com/rahafaljundi/Gradient-based-Sample-Selection
Open Datasets Yes Disjoint MNIST: MNIST dataset divided into 5 tasks based on the labels with two labels in each task. Permuted MNIST: We perform 10 unique permutations on the pixels of the MNIST images. Disjoint CIFAR-10: Similar to disjoint MNIST, the dataset is split into 5 tasks according to the labels, with two labels in each task.
Dataset Splits No The paper states training and testing details, but does not explicitly provide information about a separate validation set split (e.g., percentages, counts, or explicit mention of a validation set).
Hardware Specification No The paper does not provide specific details about the hardware (e.g., GPU models, CPU types) used for running the experiments.
Software Dependencies No The paper mentions using SGD optimizer and a neural network framework implicitly (PyTorch/TensorFlow are common in such papers, but not specified), but does not provide specific software dependencies with version numbers.
Experiment Setup Yes In all experiments, we use a fixed batch size of 10 samples and perform few iterations over a batch (1-5)... Following [10], for disjoint and permuted MNIST we use a two-layer neural network with 100 neurons each while for CIFAR-10 we use Res Net18... In all experiments, we use SGD optimizer with a learning rate of 0.05 for disjoint MNIST and permuted MNIST and 0.01 for disjoint Cifar-10.