Calibrating CNNs for Lifelong Learning
Authors: Pravendra Singh, Vinay Kumar Verma, Pratik Mazumder, Lawrence Carin, Piyush Rai
NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We perform extensive experiments on multiple benchmark datasets (SVHN, CIFAR, Image Net, and MS-Celeb), all of which show substantial improvements over state-of-the-art methods |
| Researcher Affiliation | Academia | Pravendra Singh 1, Vinay Kumar Verma 2, Pratik Mazumder1, Lawrence Carin2, Piyush Rai1 1CSE Department, IIT Kanpur, India 2Duke University, USA |
| Pseudocode | No | No pseudocode or algorithm block is present in the paper. The method is described using mathematical equations and figures. |
| Open Source Code | No | The paper does not provide any explicit statements about releasing source code for the methodology, nor does it provide a link to a code repository. |
| Open Datasets | Yes | We perform experiments on the SVHN [28], CIFAR [29], Image Net [30], and MS-Celeb-10K [31] datasets. |
| Dataset Splits | Yes | We perform experiments on the SVHN [28], CIFAR [29], Image Net [30], and MS-Celeb-10K [31] datasets. In the case of SVHN, which has 10 classes, we group 2 consecutive classes to get 5 tasks, and we incrementally train on the five tasks. ... We train the network for 150 epochs for each task with the initial learning rate equal to 0.01, and we multiply the learning rate by 0.1 at the 50,100 and 125 epochs. |
| Hardware Specification | No | The paper does not provide specific hardware details such as GPU models, CPU types, or memory specifications used for running the experiments. |
| Software Dependencies | No | The paper mentions using the 'SGD optimizer' but does not specify version numbers for any programming languages, libraries, or frameworks used (e.g., Python, PyTorch, TensorFlow, CUDA). |
| Experiment Setup | Yes | For SVHN and CIFAR-100 experiments, we use the Res Net-18 architecture. ... we train the network for 150 epochs for each task with the initial learning rate equal to 0.01, and we multiply the learning rate by 0.1 at the 50,100 and 125 epochs. ... We use the SGD optimizer in all our experiments. |