Изображение профиля frankoziga
@frankoziga
Флаг Slovenia Ljubljana, Slovenia
На сайте с 9 марта 2018 г.
0 Рекомендации

frankoziga

Онлайн Офлайн
I'm a programmer degree on the Faculty of Computer and Information Science at the University of Ljubljana. My journey started 9 years ago when I first got familiar with programming and started working on IOS and Android applications. After a few years in college, I started to learn more about optimization, machine learning and deep learning. • Tools: PyCharm, Eclipse, Sublime text, Jupyter notebook, LaTeX, Netbeans • Programming languages: Python (Advanced), Java(Advanced), Javascript, C, C++ • Machine learning libraries (for Python): Sklearn, Keras, Tensorflow • Data analysis tools: Pandas, Numpy • Web scraping: Selenium • Android applications • Machine learning algorithms: • Deep learning: Neural networks, CNN, RNN, LSTM, GRU • Text mining: NLP, word embeddings, tf-idf • Classification: NN, Logistic regression, Random Forest, SVM • Regression: SVR, Ridge regression, Lasso • Clustering: K-means, PCA, DBSCAN
$35 USD/hr
6 отзыва(-ов)
3.8
  • 100%Выполненные работы
  • 100%В рамках бюджета
  • 100%Своевременность
  • 11%Рейтинг повторного найма

Портфолио

Последние отзывы

Опыт работы

Black-jack bot with advanced card counting system

Dec 2017 - Aug 2018 (8 months)

The main idea is to build a Black-jack game and 2 models (game decision model, betting model) which we will train with Deep mind's Deep Q learning algorithm. The goal is to train game decision model which will play like a decision table that you get in a casino and betting model that will bet just like count carding expert. Game decision model shows basic playing with cards and betting model beats the game with 42% of return when 1000bets are made.

Executor - manufacture optimization

Jul 2017 - Nov 2017 (4 months)

Executor is a system of mine, which I build for our family business. Idea is to create interaction between manufacture and office in which model generates the best sequence of products, which needs to be done, for all machines. The system also allows you to track work of workers, alarms when you need to restock some product, pay stimulation, manufacture analysis...

Merlin

May 2017 - Jun 2017 (1 month)

Project Merlin I've created on my own. The task was to predict end odds of a game from a particular league to a particular bookmaker. The middle conclusion was that there was too little data because you would need end result of a games which I didn't have. We decided to continue my work and try to predict end odd based on all bookmakers. The end result was that I've managed to discover from which bookmaker bookmakers take their odds for a particular league.

Fraud detection system

Apr 2017 - May 2017 (1 month)

There was already settled project Fraud detection system for football matches which I had to improve. First, there was testing on different models and checking if there is a difference between normalizing parameters on full data and on training data. The last task was to find new features which will improve the model. The end result was that I improved model from 92% of accuracy to 95%.

Квалификация

Deep learning specialization (2018)

Coursera

The Deep Learning Specialization is designed to prepare learners to participate in the development of cutting-edge AI technology, and to understand the capability, the challenges, and the consequences of the rise of deep learning. Through ve interconnected courses, learners develop a profound knowledge of the hottest AI algorithms, mastering deep learning from its foundations (neural networks) to its industry applications (Computer Vision, Natural Language Processing, Speech Recognition, etc.).

Верификации

  • Соединить с Facebook
  • Предпочтительный фрилансер
  • Оплата подтверждена
  • Телефон верифицирован
  • Личность подтверждена
  • Электронный адрес подтвержден

Мои топ-навыки

Искать похожих фрилансеров