yjaenike Goto Github PK

bachelor-s-thesis-code

The code used in my Bachelor's Thesis

cv-tricks.com

Repository for all the tutorials and codes shared at cv-tricks.com

databaselogrec

Implementation of a simple logging and recovery procedure for a simulated database

dino-ml

🦎 Simple AI to teach Google Chrome's offline dino to jump obstacles

gfpgan

GFPGAN aims at developing Practical Algorithms for Real-world Face Restoration.

gst

Generalised Suffix Tree implementation

hacker-news-salaries-data

Data exploration of software developer salary info from Hacker News

hide-my.io

icp-master-thesis

igd2021

The Interactive Game Development Project of 2021

image-classifier

Simple image classifier from: https://nikhilsrambles.substack.com/p/how-to-build-an-insanely-good-image-classifier-in-under-10-minutes

lectures

This is the repository for all the lectures. As github is only for sourcecode, I'll only link larger stuff like videos. This repository will probably be more up-to-date than the corresponding Stud.IP files

light_cycle

A Jekyll theme for automatically generating and deploying landing page sites for mobile apps.

linesearch

A java command line tool to find a string in a document.

linux_hot_corner_suspend

Shell script that can be run as a "Hot Corner" command to get into Sleep Mode

master-thesis

In my thesis I want to investigate the effectiveness of different manifold manipulation techniques on the performance of transformer based time series forecasting methods. Transformer models can work with time series , however, those sequence-to-sequence models rely on neural attention between timesteps, which allows for temporal learning but fails to consider distinct spatial relationships between variables inside the timestep. To solve this problem, Jeng et al. (2022) have introduced a new embedding methodology (Spacetimeformer) to capture the spatiotemporal relationship between variables. As part of my thesis, I want to compare 4 different models: Transformer Transformer + Manifold Mixup Spacetimeformer Spacetimeformer + Manifold Mixup Since time-series data is not easily interpretable by humans, I will use PCA and t-SNE to map the multi-dimensional output sequence vectors into two dimensions to visually observe the similarity in the distribution of the synthetic data and real data instances.

node-cobol

:tv: COBOL bridge for NodeJS which allows you to run COBOL code from NodeJS.

procedure

https://yjaenike.github.io/procedure/

swiftdownnook

📦 A themable markdown editor component for your SwiftUI apps.

test