Group of Horribly Optimistic STatisticians

Welcome to GHOST! Our science club is a community of enthusiastic data scientists who are passionate about exploring the world through numbers and probabilities. Whether you are a seasoned data professional or just starting to explore the world of statistics, join us and become a part of the next generation of optimistic statisticians!

Join usLearn more
That's us

That's us

From the left: Łukasz, Piotrek, Agata, Antoni,  Sebastian, Krzysztof, Adam, Konstanty, Asia, Przemek, Zuzanna, Oskar, Jacek, Alex, Michał, Ignacy, Anna,  Łukasz, Tymon, Wojciech, Alicja,  Zuzanna, Kacper (behind the camera) and many more that preferred coding instead of photo shoot.

We meet each week and do fun stuff with not only Machine Learning.

Our sections

Sections are conducted in the Polish language. However, if there are people from abroad willing to join, we change to English. Schedule of meetings will be revealed soon.

Applied Intro To ML

The continuation of Intro is ML from the first semester in a practical version. Within the section, members will be able to jointly undertake smaller ML projects or competitions on the Kaggle platform in order to extend their competences or expand their portoflio. At meetings, we will discuss the projects undertaken, solve problems encountered, and work together with various learning materials.

Quantum Computing for ML

For those eager to stay in the loop with the bleeding (if not yet-to-bleed) edge of computer science, quantum computing might be the way to go. With its undeniable potential and numerous disruptive feats, the intersection of quantum mechanics and computer science spurs as much excitement in both the academia and industry as few other topics. In this section, we will strive to master exactly the juicy bits of theoretical physics and ML which make it the case. As rewarding as it will be challenging, our shared journey through the intricacies of quantum mechanics, mind-boggling algorithms, reimagined computing, and way more, will certainly be an enriching experience. So, if brain-tickling concepts are right up your street, if you want to broaden your ML skillset, or you simply don’t want your algebra and calculus semesters to go in vain, joining our section of QTs (Quantum Tinkerers) will be an excellent choice! After all, everyone can agree that all the most enjoyable aspects of AI always start with a tad of theoretical physics and highly abstract mathematics ;)

Ghost Day comitee

The GHOST Day Conference Committee is a close-knit group of extraordinary individuals dedicated to organizing one of the most prestigious scientific events in the field of machine learning. Our main mission is to create an excellent platform for machine learning fanatics to share their experiences and knowledge, and to provide an inspiring place to update their skills in a rapidly changing field.
We enjoy the proud support of renowned experts in machine learning and committed corporate partners, which has already allowed us to organize four exceptional editions of our conference. Positive feedback from participants, speakers and partners is our best recommendation for the quality of our activities.
Our dedicated committee team is constantly working to provide attractive speakers and topics for future editions of the conference. For machine learning enthusiasts looking to stay up to date with the latest trends and developments, there is no better place in Poznan than GHOST Day. This is a unique opportunity to learn, get inspired and network with like-minded people.

GHOST 229

The section was inspired by the CS 229 course from Stanford, and it is the material from this course that we will build on to create a good foundation in Machine Learning. Once we have learned some new theoretical concept, we will try to implement it for better understanding. The topics covered in the section will be quite similar to those in the book "Introduction to Statistical Learning", but they will be covered from a different point of view so if someone has attended the Intro to ML section then they will also benefit from attending GHOST 229 meetings. No knowledge of ML is required to attend the section, but elements of Linear Algebra and Calculus II may appear but a lack of prior knowledge of these will not make it too difficult to understand the topics covered in the section.

RL in Games

A section for lovers of games and artificial intelligence. With us you will learn the theoretical foundations of reinforcement learning and how to apply your knowledge in practice by participating in exciting projects. You will learn how to teach AI to play DOOM, chess or even GO. The class is suitable for people of all levels but a good knowledge of Python will be useful. You are welcome!

Machune Unlearning (MU) Seminar

Recent AI regulations, such as the "right to be forgotten," mandate the removal of private user information from both computer systems and ML models upon request. While deleting data from backend databases seems straightforward, it's inadequate in the AI context because ML models often retain old data. This highlights the need for a new approach, known as machine unlearning, to help ML models forget specific data and concepts.In this section we will dive into the world of Machine Unlearning starting from beginning. We will start by defining the term and reading survey papers, then, we'll delve into state-of-the-art techniques, focusing on deep learning applications. Good understanding of classical Machine Learning techniques and Deep Learning is preferred, but no prior MU experience is required. We welcome all foreign students and will be happy to organize meetings in English!

Neurosymbolic AI

Neurosymbolic AI is a very rapidly growing branch of the field of artificial intelligence. It addresses the integration of symbolic methods such as programs, data structures and symbolic solvers with methods known from deep learning such as transformers or graph networks. This allows for formally verifiable, interpretable and prior knowledge-based solutions, which is key in many areas such as robotics, theorem proving or scientific research support. At the meetings, we plan to both discuss scientific articles and discuss practical implementations.

Language & Planning Hub

We are a team of enthusiasts where the world of natural language processing (NLP) meets the fascinating field of personal information management - PIMPS (Personal Information Management and Planning Systems). Are you a person who: Loves to put the power of language into practice? Is interested in the latest trends in NLP and maybe even fascinated by RAG (retrieval augmented generation)? Believes that language is not just a tool, but the best interface for understanding the world? Would you like to experiment with using NLP tools to automate personal data management? Maybe you are passionate about note-taking and information gathering? Then it is for you that we have created the NLP-PIMPS Hub! This is where we come together to share ideas, explore the theoretical aspects of NLP while using the latest tools to effectively manage personal data. Whether you are more of a theoretician or a practitioner, our meetings are the place where these two fields come together harmoniously. Join us to be part of an innovative community where science, practice and creativity meet in fascinating ways. NLP-PIMPS Hub - where we are creating the future of linguistic automation and effective personal data management!

Computer Vision Seminar

Computer Vision Seminar is the section where we learn about the methods of computer vision. In our second semester since reactivation, we are further completing the necessary basics so that we will soon be dealing with the latest publications in the field. Last semester we were on a trip - in December we visited the VWP factory in Września and saw the applications of CV in car production. Is it worth joining us? Yes! Meetings are held once a week and usually last less than an hour.

Tiny ML

TinyML is a construction-oriented section where we work on projects based on microcomputers and using machine learning solutions. In the upcoming semester, we will continue working on our six-legged robot based on the Jetson Nano platform. We will develop the control system for our robot and prepare image recognition models.

Found your favourite section already?

Contact

We also invite people who would like to present some data analysis topics/projects at one of our meetings. We are open to suggestions!

Email :

ghost@put.poznan.pl

Address :

Institute of Computing Science, ul. Piotrowo 2, 60-965 Poznan, Poland