Neelkanth Pradhumanbhai Patel

Title: Neelkanth Pradhumanbhai Patel: Inventor in Machine Learning Input Scaling.

Introduction

Neelkanth Pradhumanbhai Patel is an inventor based in Scarborough, Canada, associated with Qualcomm Incorporated. His patent work concerns efficient input scaling for machine-learning models.

Latest Patents

Patel is credited with 1 patent, titled “Input Scaling with Convolutional Layers.” The invention describes techniques for scaling an input across one or more dimensions before a machine-learning model processes it. It determines a stride length for each scaled dimension from the input’s starting size and the target size, and at least one stride length may be non-integer. The model then generates an inference using the scaled input.

Career Highlights

Patel’s listed patent reflects work on preparing inputs for machine-learning inference. The disclosed approach addresses scaling across multiple dimensions, including cases where fractional stride lengths are used.

Collaborations

Patel’s named coworkers include Haoping Xu and Prajakt Kulkarni. The patent information provided identifies Qualcomm Incorporated as his company.

Conclusion

Neelkanth Pradhumanbhai Patel’s patent highlights an approach to efficient input scaling for machine-learning models, developed in connection with Qualcomm Incorporated.