Data Science Meetup

Join us at new Data Science Meetup! We have prepared several reports, will be interesting and useful! 🔹 Speech #1: Rustem Galiullin, AK BARS Digital Technologies Topic: "Scrum at Data Science". 1. How to deliver a model within 2 weeks using machine learning cycle? 2. User story decomposition by CRISP-DM 3. Sprint review is a vital part to improve code readability and knowledge exchange among data scientists 4. Sprint is a time boundary for model development 5. Product backlog and sprints are the way to increase your value for other divisions 🔹 Speech #2: Bulat Lutfullin, Provectus Topic: "ML for content generation. How to generate a level for Hotline Miami 2". 1. Pure random is still used in gamedev, while ML is able to generate tons of content 2. What is a genetic algorithm? How can it be applied for the level generation? 3. Are there other machine learning methods for content generation? 4. Do we really need ML in gamedev? 🔹 Speech #3: Rinat Gareev, Provectus Topic: "Scammer Detection on Crowdsourcing Platforms". Microtask crowdsourcing is a trade-off between budget, the speed of getting results and their quality. It is a popular approach to quickly build a training dataset for machine learning. There are standard quality assurance mechanics on crowdsourcing platforms but they still might leave an opportunity for scamming. In this presentation, I will talk about our experience in improving scammer detection, how to estimate fairness and proficiency of contributors using unsupervised machine learning models ✔ Language: Russian When? 18 March, 10:30 am Where? GrowUP, Ayvazovskogo, 3 Free admission, registration is required: 👉 https://provectus.com/events/#data-science-kazan Still have questions? Please, write us: tkovaleva@provectus.com