Nadchádzajúce meetupy
ZRUŠENO
Why your own neural network implementation can be faster than any framework? The secret is when. (Rudolf Jaksa)
- Language: Slovak
- Date and time: ZRUŠENO
Wednesday, March 11th at 16:30
Naive implementation of neural networks in the C language can be tens, or hundreds times faster than usage of common neural networks frameworks. Simplicity of neural networks learning algorithms goes well with the automatic vectorization in recent versions of GNU C Compiler and in the result native builds are much faster than generic, and newer CPUs are significantly faster than just few years old ones. We will show possible gains. Concrete speedup depends on the topology of the network and on the CPU itself. To measure it, we prefer the classic Connection updates per second (CUPS) training speed metric, which is more universal and more practical than task-specific benchmarks. “To measure” is important, as the interplay of topology, framework and hardware is complex. Our C code for neural networks training and support tools for automation of the CUPS curves measurement are on GitHub.
Rudo is the Head of AI in Matsuko, working on custom 3D convolutional architectures for holographic communication. Before he worked on NN predictors and RL for E-commerce in Exponea, and on the industrial and weather NN prediction in Kybernetes/MDJ. For 20 years he taught Neural Networks on the TU Kosice. He worked with Interactive Evolutionary Computation on KID Fukuoka.
Minulé meetupy
Optimization for ML: From Theory to Practice and Back (Filip Hanzely)
- Language: English
- Date and time: Tuesday, April 21th at 18:00
- Venue: online (see meetup.com)
- Social: Facebook, Meetup.com, DAM, Eventbride
Most of supervised machine learning problems, including deep learning, are routinely solved via optimization recently. We will discuss several widely used optimization algorithms and mention how the current theory is (not) reflected in the practice. Lastly, we will talk about several challenges the field is currently facing.
Filip Hanzely is now PhD Student (Optimization) on KAUST. He received his Bc degree (Economics and Financial mathematics) at the Comenius University, Bratislava, Slovak Republic in 2016, and MSc degree (Mathematics and Statistics) at the University of Edinburgh in 2017. He had interned in Amazon, Berlin and in Google Research, New York, in summer 2018 and summer 2019, respectively.
Formal concept analysis (FCA) and its basic tools (Ondrej Krídlo)
- Language: Slovak
- Date and time: Wednesday, February 26th at 16:00
- Venue: VKM room, Jesenná 5, Košice, first floor
- Social: Facebook, Eventbride, Meetup.com
During the lecture we will explain the fundamental concepts and approaches of FCA. The method was created in the 80’s with the intention to make the object-property datatype analysis results reflect reality with greater accuracy. This type of analysis enables us to see previously hidden meanings and relations in object clusters in regards to their attributes. Fundamental principles of FCA are built for the true/false datatype. We will showcase the simple transition to the analysis of more sundry datatypes.
RNDr. Ondrej Krídlo, PhD. studied Informatics at Faculty of Science, P. J. Šafárik University in Košice, where he currently works. In collaboration with a team in Malaga, he focuses on formal concept analysis, fuzzy logic and category theory.
Text detection and recognition in natural scene images – a real challenge (Peter Bugata, Dávid Hudák)
- Language: Slovak
- Date and time: Wednesday, February 12th at 16:00
- Venue: VKM room, Jesenná 5, Košice, first floor
- Social: Meetup.com, Facebook, Eventbrite
During the summer student internship at VSL Software, a.s., we have learned some text recognition techniques. We applied them to the practical problem of recognizing the wagon number in freight trains. We will present our results achieved through two neural networks that have learned to cooperate.
Peter Bugata is a project manager at VSL Software, a.s. in Košice. He studied Teoretická kybernetika, matematická informatika a teória systémov at Faculty of Science, P. J. Šafárik University in Košice.
Introduction to Deep Reinforcement Learning II (Slavo Maťašovský)
- Language: slovak
- Date and time: Tuesday, November 26th at 18:00
- Venue: UVP Technicom, Hviezdoslavova 7, Košice
workshop description:
1) From Reinforcement Learning to Deep Reinforcement Learning
2) Value Based Methods
– Deep Q-Learning Agent
– Dueling Deep Q-Learning Agent
3) Policy Gradient Methods
– Introduction to PG methods – state action-value and policy parameterization using DNN – Deep Neural Networks
– Policy Gradient Theorem
– A2C – Advantage Actor-Critic Agents
– DDPG – Deep Deterministic Policy Gradient Agent
– MADDPG – Multi-Agent Deep Deterministic Policy Gradient
– SAC – Soft Actor Critic Agent
4) Stabilizing Deep Reinforcement Learning
– Experience Replay
– Fixed Q-Targets
5) Examples of RL agents using ML Agents environment
– Car simulator using SAC/DDPG
– Table tennis multiplayer game using MADDPG
– Robotic arm object fetching
Slavo Matašovský vyštudoval inžinierský titul v odbore computer science na Technickej univerzite v Košiciach. Aktuálne pracuje vo firme Tachyum (www.tachyum.com) ako senior AI Architect.
Introduction to Deep Reinforcement Learning (Slavo Maťašovský)
- Language: slovak
- Date and time: Thursday, May 30th at 18:00
- Venue: Eastcubator, Hviezdoslavova 7, Košice
Reinforcement Learning je efektívna metóda ako naučiť vlastného inteligentného autonomného agenta riešiť rôzne AI problémy – hry, samostatne-riadené vozidlá, robotiku, šetrenie energie v dátovych centrách a pod. Proces učenia závisi na hodnote odmien a na pozorovaniach získanými interakciami autonomného agenta s prostredím. Tento model je veľmi všeobecný a možno ho použiť v mnohých praktických situáciách. Prejdeme si hlavne metody DRL – Q-Learning a Policy Gradients, vysvetlíme si aké sú výhody kombinovania Deep Neural Networks s DRL a nakoniec si ukážeme zopár príkladov bežiacich v simulačnom prostredí Unity ML-Agents.
Slavo Matašovský vyštudoval inžinierský titul v odbore computer science na Technickej univerzite v Košiciach. Aktuálne pracuje vo firme Tachyum (www.tachyum.com) ako senior AI Architect.
Automation of Data Science (Tomas Horvath)
- Language: slovak
- Date and time: Tuesday, February 25th at 18:00
- Venue: Eastcubator, Hviezdoslavova 7, Košice
With the crescent number of data scientific tasks, there is a high demand for the use of machine learning techniques to extract new and meaningful knowledge from new datasets. The choice of best techniques and adequate setting of their hyper-parameters are the main subject of a research area known as automated machine learning, often abbreviated as AutoML. During the talk, the main methods used and utilized to simulate an expertise and experience of a data scientist as well as techniques for automation of certain tasks in data science will be presented. The talk is rather theoretical, basic knowledge of machine learning and data mining is eligible to understand the basic concepts behind AutoML.
Dr. Tomáš Horváth is the head of the newly established Data Science and Engineering Department at the Eötvös Loránd University in Budapest, Hungary, since September 2016. Here received his MSc and PhD degrees at the Pavol Jozef Šafárik University in Košice, Slovak Republic, in 2002 and 2008, respectively. He was on a post-doc internship at the Information Systems and Machine Learning lab of the University in Hildesheim, Germany, from 2009 to 2012. From 2015 to 2016 he received a post-doc grant at the Department of Computer Science, University of São Paulo in São Carlos, Brazil.
His research interests include relational learning, rule-based and monotone classification techniques, pattern mining, recommender systems and personalization. Recently, he is focusing his work on meta-learning techniques and automated machine learning approaches.
AI a AR v robotike (Jozef Orenič)
- Language: slovak
- Date and time: Tuesday, January 29th at 18:00
- Venue: Eastcubator, Hviezdoslavova 7, Košice
Ako integrovať AI a AR do robotiky? Aké sú dnešné možnosti na rozdiel od minulosti? Uvedieme si niekoľko príkladov, čo sa v súčasnosti v robotike robí, čo by sa v blízkej budúcnosti dalo a čo je pre najbližšie roky stále len fikcia. Vďaka veľkému rozmachu a investícií do aplikovania umelej inteligencie v reálnych a užitočných situáciach, je tento pojem posledné roky skloňovaný v takmer každom odvetví. Čo s týmito technológiami môžu dosiahnuť jednotlivci či samoukovia? Pokúsim sa to objasniť na príklade spojenia robotiky a rozšírenej reality. Nebude chýbať aj malá ukážka aktuálne zostrojovaného robota práve s týmto cieľom.
Jozef Orenič vyštudoval inžinierský titul v odbore informatika na Technickej univerzite v Košiciach, no už roky predtým sa venoval vývoju webových stránok a systémom. Popri tom ho ale stále zaujímala robotika, v ktorej neustále využíval svoje vedomosti z rôznych programovacích jazykov. Robotika je pre neho stále hobby, no rád by sa tomu venoval naplno. Aktuálne pracuje vo firme GlobalLogic ako team leader, kde sú jeho najväčšie výzvy vedenie ľudí k lepšiemu programovaniu a ešte lepšej tímovej nálade.
Introduction to H2O Driverless AI with Jan Gamec
- Language: slovak
- Date and time: Wednesday, December 19th at 18:00
- Venue: Eastcubator, Hviezdoslavova 7, Košice
H2O.ai is a visionary Silicon Valley open source software company that created and reimagined what is possible. We are a company of makers that brought to market new platforms and technologies to drive the AI movement. We are the makers of H2O, the leading open source data science and machine learning platform used by nearly half of the Fortune 500 and trusted by over 14,000 organizations and hundreds of thousands of data scientists around the world.
H2O Driverless AI employs the techniques of expert data scientists in an easy to use application that helps scale your data science efforts. Driverless AI empowers data scientists to work on projects faster using automation and state-of-the-art computing power from GPUs to accomplish tasks in minutes that used to take months.
With Driverless AI, everyone including expert and junior data scientists, domain scientists, and data engineers can develop trusted machine learning models. This next-generation automatic machine learning platform delivers unique and advanced functionality for automatic data visualization, feature engineering, model interpretability and low-latency deployment.
Jan is Senior SW/ML Engineer at H2O.ai working mostly on Driverless AI. In past Jan worked as full-stack engineer on various projects oriented on ML, web/mobile apps or cryptography, and holds Master’s degree in ML from Prague CTU. Out of office, Jan is keen on sports and playing violin.
Apache Mesos – ako sa budujú moderné clustre (Zdenko Vrábel)
- Language: slovak
- Date and time: Wednesday, October 17th at 18:00
- Venue: Eastcubator, Hviezdoslavova 7, Košice
Zdenko je 16 rokov programátor a nič iné len prográmator so záľubou pre infraštruktúry a netradičné prístupy.
Save the Vineyard Hackathon – Preparatory Workshop (Stanislav Hrivňák)
- Language: slovak
- Date and time: Tuesday, September 4th at 18:00
- Venue: DvaDve, Kuzmányho 3, Košice
Workshop je vhodný pre začiatočníkov aj stredne pokročilých v oblasti strojového učenia.
RNDr. Stanislav Hrivňák pôsobí na UPJŠ v Košiciach ako počítačový biofyzik, kde sa dlhodobo venuje rozvoju algoritmov na spracovanie dát z mikroskopov. Takisto je aj súčasťou tímu Machine Intelligence Lab v GlobalLogic, kde aktívne spolupracuje na vytváraní machine learning modelov.
Practical TensorFlow – Wide and Deep Learning Workshop (Daniel Kuchta)
- Language: slovak
- Date and time: Thursday, May 31st at 17:30
- Venue: BCK, Štúrová 27, Košice
Na workshop si prosím prineste vlastné notebooky. Počet miest je limitovaný na 15 účastníkov. Workshop je vhodný aj pre začiatočníkov.
Postavte si svoje AI riešenie (Peter Bednár)
- Language: slovak
- Date and time: Wednesday, May 9th at 18:00
- Venue: Eastcubator, Hviezdoslavova 7, Košice
Building Safe AI (Andrew Trask)
- Language: slovak
- Date and time: Tuesday, April 10th at 18:30
- Venue: Eastcubator, Hviezdoslavova 7, Košice
In the second half of this talk, I’ll be discussing the significant impacts this technology has when combined with the recent advancements in Blockchain and Peer-to-Peer into a new open-source platform called OpenMined. This will include a live demo showing how to train a neural network on a large, distributed, private dataset.
Deep learning: čo, prečo, ako? (Rudolf Jakša)
- Language: slovak
- Date and time: Tuesday, February 20th at 18:00
- Venue: Eastcubator, Hviezdoslavova 7, Košice
na báze umelej inteligencie. Aktuálne pracuje na vývoji predikčných a modelovacích aplikácii postavených na neurónových sieťach v Kybernetes s.r.o. Košice. Dlhodobo sa venuje algoritmu Backpropogation a otázke implementácie neurónových sietí a práci s numerickými dátami.
Rozpoznávanie šiat vo fashion industry – neurónky vs. geometrický prístup (RNDr. Andrej Hosťovecký, PhD.)
- Language: slovak
- Date and time: Tuesday, January 16th at 18:00
- Venue: Eastcubator, Hviezdoslavova 7, Košice
Rankingová Alenka ve fintechové říši divů (Marek Modrý)
- Language: czech
- Date and time: Tuesday, December 5th at 18:00
- Venue: Eastcubator, Hviezdoslavova 7, Košice
Live predictions with schemaless data at scale (Ondrej Brichta)
- Language: slovak
- Date and time: Tuesday, November 7th at 18:00
- Venue: Eastcubator, Hviezdoslavova 7, Košice
We will show you, how this is done at Exponea and much more. How to connect this data to Spark ML library and then translate the model into a sequence of mathematical functions and aggregation methods for our in memory database to evaluate it on all customers in real time. Ondrej Brichta – currently working at Exponea as AI Engineer. Studying Logic and computability at Vienna University of Technology, alumni of Nexteria Leadership Academy and Matfyz in Bratislava
Klasifikácia rakovinových nádorov pomocou machine learningu (Daniel Kuchta)
- Language: slovenský
- Date and time: utorok 3. október o 18:00
- Venue: Eastcubator, Hviezdoslavova 7, Košice
Máš nápad na prednášku? Ozvi sa nám na info@mlmu.sk