PyData Eindhoven 2021
2021-11-12
Eindhoven, Netherlands
[PyData Eindhoven 2019, PyData Eindhoven 2020, PyData Eindhoven 2021, PyData Eindhoven 2022, PyData Eindhoven 2023, PyData Eindhoven 2024, PyData Eindhoven 2025]
| title | author | length |
|---|---|---|
| Combining Imitation + Reinforcement Learning to win the Bot Bowl competition | Nicola Pezzotti | 29m |
A hybrid agent called MimicBot combines behavioral cloning from a scripted Fantasy Football AI with reinforcement learning updates to stabilize training and… | ||
| The journey from home-brew analysis scripts to a product | Andrew Rutgers | 32m |
Starts with a working Python/Matlab analysis and turns it into a SaaS that actually runs at scale: a simulation engine packaged for customers as ChargeSim with… | ||
| Some Attention for Attenuation Bias | Ruben Mak | 31m |
The talk covers attenuation bias (measurement error/regression dilution) that shrinks model coefficients toward zero when features are noisy, and explains… | ||
| Be kind to yourself! Spend (less) time on Data Exploration | Willem Hendriks | 30m |
Argues that careful exploration still pays back double and shows concrete results on the California housing dataset: house age and price are clipped at 50… | ||
| How to quickly build Data Pipelines for Data Scientists | Geert Jongen | 31m |
Claim: build runnable data pipelines fast using pyspark and Delta Lake with notebook-driven flows. Implementation details: use streaming reads with triggerOnce… | ||
| An incomplete list of implementing data science effectively | Erick Webbe | 31m |
Claims a practical six‑item playbook and shows it in action: a sewage‑to‑hospitalization forecast that trades raw infection signal for actionable hospital bed… | ||
| MLOPS 101: Tips, Tricks and Best Practices | Vladimir Osin | 34m |
A practical checklist replaces hype with actions scoping asks the questions that kill 80% of ML projects at scoping by forcing clear success metrics and data… | ||
| Synergizing model and human decision-making using augmented machine learning | Kay Hoogland | 34m |
Models predict underwriter decisions rather than defaults, feature values and descriptions appear in the dashboard, and per-decision SHAP explanations show… | ||
| When performance matters: running a real-time computer vision pipeline | Jelmer Wilhelm | 30m |
A real pipeline claims a hard 300ms end to end and shows what actually meets that bar: OpenCV for capture and preprocessing, YOLO models for player and ball… | ||
| Biosensor Machine Learning with Julia | Matthijs Cox | 26m |
Describes extracting EMG from consumer biosensors via the open source BrainFlow, processing and training in Julia, and deploying a saved model with JLD2 to do… | ||
| Compressive Sensing | Iman Mossavat | 29m |
Compressed sensing claims that fewer measurements can replace dense sensing by moving complexity into algorithms and reconstruction, with concrete wins like… | ||
| Optimal on Paper, Broken in Reality | Vincent D Warmerdam | 37m |
Shows a concrete method to temper optimistic benchmarks by plotting accuracy versus training size and exposing how grid-search numbers hide sample-size… | ||
| Dealing with the versioning of production-ready models | Corné Vriends | 33m |
Company runs Toon® on Databricks and Spark, serving predictions at scale and tracking ~500 models with eight active deployments while processing ~1 TB/day and… | ||
| Applied AI on the edge | Maurits Kaptein | 36m |
Many PoCs never reach production because streaming latency, cloud cost, privacy and unreliable networks block deployment. A usable path compiles ONNX model… | ||
| Responsible implementation of Machine Learning models for life-and-death decision making | Michele Tonutti | 35m |
Pacmed Critical claims responsible ICU decisions by combining concrete safeguards: business rules and out‑of‑domain detection block predictions for… | ||
| Explainable causal inference results | Thomas Nägele | 33m |
Claims a graph-only algorithm — the extended Markov Blanket — finds which observable variables actually matter for a node’s posterior, implemented using a C++… | ||
| Estimating solar production w/ distributed deep learning from smart meter data | Rik van der Vlist, Corné Vriends, Fenno Vermeij | 37m |
Claim: estimate per‑interval household solar production from smart‑meter redelivery alone, using day‑level inputs of 96 per‑day 15‑minute intervals… | ||
| Real-time Transaction Categorization w/ Bayesian Feedback Loop | Tijl Kindt | 39m |
ING built a production system to categorize customers’ bank transactions in real time using a batch/streaming architecture: PySpark batch jobs collect implicit… | ||
| nothing matches | ||