Search Constraints

Start Over
Filtering by: Creator Merrifield, Matthew S. Remove constraint Creator: Merrifield, Matthew S. Contributor Agostinelli, Victor Remove constraint Contributor: Agostinelli, Victor Degree Level Master's Remove constraint Degree Level: Master's

Search Results

  • Partitioning-Guided K-Means for Effective Empty Cluster Resolution in Extreme Model Compression

    Creator:
    Huang, Tianhong
    Abstract:
    Compactness in deep learning can be critical to a model’s viability in low-resource applications, and a common approach to extreme model compression is quantization. We consider Iterative Product Quantization (iPQ) with Quant-Noise [Fan et al., 2020] to be state-of-the-art in this area, but this quantization framework suffers from preventable inference...
    Resource Type:
    Masters Thesis
    Full Text:
    across a specified span factor s and evenly spaced [Gupta et al., 2015, Courbariaux et al., 2015]. A bias