liujihai's starred repositories
pytorch-cpp
C++ Implementation of PyTorch Tutorials for Everyone
tf-encrypted
A Framework for Encrypted Machine Learning in TensorFlow
riscv-v-spec
Working draft of the proposed RISC-V V vector extension
Literatures-on-Homomorphic-Encryption
A reading list for homomorphic encryption
FHE-MP-CNN
Implementation of deep ResNet model on CKKS scheme in Microsoft SEAL library using multiplexed parallel convolution
LibTorch-ResNet-CIFAR
ResNet Implementation, Training, and Inference Using LibTorch C++ API
TrainableActivation
Implementation for the article "Trainable Activations for Image Classification"
openfhe-logreg-training-examples
OpenFHE-Based Examples of Logistic Regression Training using Nesterov Accelerated Gradient Descent
fabric-crosschain
Fabric crosschain research
function-approximation
Approximating a sine function with polynomials using Tensorflow, demonstrating how to use weights, SSE loss function, gradient descent, and backpropagation
Approximating-Continuous-Function-With-ReLU
ReLU's Function Approximation
Exploring_Neural_Network
A quartic function approximator was built using Tensorflow and Keras API to approximate how ReLU, tanh, and sigmoid activation functions behave using shuffled and unshuffled dataset using two different network structures. Also, a multilayer neural network was designed to approximate XOR function encountered in digital logic.
minimaxApprox
:exclamation: This is a read-only mirror of the CRAN R package repository. minimaxApprox — Implementation of Remez Algorithm for Polynomial and Rational Function Approximation. Homepage: https://github.com/aadler/MiniMaxApprox Report bugs for this package: https://github.com/aadler/MiniMaxApprox/issues
PolynomialLR_withoutLibrary
Polynomial linear regression: First generate dataset using sine function and add Gaussian noise with mean zero and variance 0,1. Consider the polynomial equation of degree k in x having parameters β0, β1 ... ... . βk. Derive the loss function for each k-degree and optimize it using gradient descent.
ML_project
Project that consists of approximating a function using neural networks with varying numbers of layers in the set L=[2,4,8,16,32,64,128] and also varying numbers of neurons in the set N=[2,4,8,16,32,64,128,256]. All these were implemented for 3 different activations of hidden layers: ReLU, Tanh, Sigmoid.
Parameter-optimization-for-DeepZ-robustness-verification
DeepZ is a method for local robustness verification, based on zonotopes, a type of convex relaxations, and abstract transformations. While affine transformations and convolutions can be represented exactly using this framework, a sound over-approximation has to be used to approximate the ReLU function. The abstract transformer used in DeepZ is parameterized by one parameter per hidden unit. In this work we propose an optimization strategy for this parameterization to increase the maximum verifiable image perturbation for both fully-connected and convolutional network architectures.