Abstract: According to examples, a system for implementing dedicated feature-based techniques to optimize inference performance in a neural network may include a processor and a memory storing instructions. The processor, when executing the instructions, may cause the system to determine enhanced relationship and interaction information associated with the first object and the second object and an initial feature value and generate a modified first embedding and a modified second embedding. The processor, when executing the instructions, may further cause the system to determine an updated feature value utilizing the modified first embedding and the modified second embedding and generate a prediction and a prediction loss associated with the first object and the second object utilizing the modified first embedding and the modified second embedding.

Abstract: A method of providing an adaptive bit-width neural network model on a computing device, comprising: obtaining a first neural network model, wherein each layer of first neural network model has a respective set of parameters expressed with an original bit-width of the first neural network model; reducing a footprint of the first neural network model by using respective reduced bit-widths for storing the respective sets of parameters of different layers of the first neural network model, wherein: preferred values of the respective reduced bit-widths are determined through multiple iterations of forward propagation through the first neural network model using a validation data set while each of two or more layers of the first neural network model is expressed with different degrees of quantization until a predefined information loss threshold is met; and generating a reduced neural network model with quantized parameters expressed with the respective reduced bit-widths.