Bijay Gaudel's Projects
Pytorch Implementation of different types of AutoEncoders
DeepWalk is network embedding technique proposed for learning the representations of nodes in a network, which is able to preserve the neighborhood structure of the nodes within a short random walk.
A diffusion-based denoising approach to mitigate online adversarial image attacks and an FFT-based detector
Simple django blog post site.
Face detection using webcame
Implementation of different types of GAN in pytorch
Config files for my GitHub profile.
Local and Global similarity calculation
Line is a graph embedding technique, suitable for arbitrary types of information networks: undirected, directed, and/or weighted. The method optimizes a carefully designed objective function that preserves both the local and global network structures.
A general framework that leverages node feature information (e.g., text attributes) to efficiently generate node embeddings for previously unseen data. Instead of training individual embeddings for each node, GraphSAGE learn a function that generates embeddings by sampling and aggregating features from a nodeβs local neighborhood. Here, the implementation of GraphSAGE is based on transductive training
Graph sampling and attention
landing gear detection training using YOLOV4.
NepaliAutoCompleteML is a deep learning framework (using Pytorch) for suggesting relevant words during writing. We used Recurrent Neural Network with layer GRU or LSTM, and KL divergence as loss function to train our model.
Node2Vec is an effective graph embedding technique which learns a mapping of nodes to a low dimensional space of features that maximizes the likelihood of preserving network neighborhoods of nodes.
Structural Deep Network Embedding (SDNE) is a semi-supervised deep model for graph embedding. Which has multiple layers of non-linear functions, thereby being able to capture the highly non-linear network structure.
Simple Flask API and Flutter App to track sensors data
Simple neural network visualiser from coursera project
Soft Nearest Neighbor loss from paper (https://arxiv.org/pdf/1902.01889.pdf) Implemented on tf 2.x
Structural identity is a concept of symmetry in which network nodes are identified according to the network structure and their relationship to other nodes. Structural identity has been studied in theory and practice over the past decades, but only recently has it been adressed with representation learning techniques. Node2vec is a flexible framework for learning latent representation for the structural identity of nodes. Struc2vec uses a hierarchy to measure node similarity at different scales, and constructs a multilayer graph to encode structural similarities and generate structural context for nodes.
Here I have coded the random_walk and biased random walk on the graph.
Basic of Deep Learning, Neural Networks, TensorFlow, and Machine Vision