A Walsh Hadamard Derived Linear Vector Symbolic Architecture

Authors: Mohammad Mahmudul Alam, Alexander Oberle, Edward Raff, Stella Biderman, Tim Oates, James Holt

NeurIPS 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental section 4 will empirically evaluate HLB in classical VSA benchmark tasks, and in two recent deep learning tasks, showing improved performance in each scenario.
Researcher Affiliation Collaboration 1University of Maryland, Baltimore County,2Booz Allen Hamilton, 3 Laboratory for Physical Sciences
Pseudocode No No pseudocode or algorithm blocks were explicitly labeled or presented in a structured format.
Open Source Code Yes Code is available at https://github.com/Future Computing4AI/ Hadamard-derived-Linear-Binding.
Open Datasets Yes CSPS experimented with 5 datasets MNIST, SVHN, CIFAR-10 (CR10), CIFAR-100 (CR100), and Mini-Image Net (MIN). ... The network is trained on 8 datasets listed in Table 4 from [4]
Dataset Splits Yes Other than changing the VSA used, we follow the same training, testing, architecture size, and validation procedure of [3].
Hardware Specification Yes All the experiments are performed on a single NVIDIA TESLA PH402 GPU with 32GB memory.
Software Dependencies No The paper mentions 'The Torch HD library [15] is used for implementations of prior methods.' but does not provide specific version numbers for any software dependencies.
Experiment Setup Yes Other than changing the VSA used, we follow the same training, testing, architecture size, and validation procedure of [3].