PyData Berlin 2018
2018-07-06..2018-07-08
Berlin, Germany
[PyData Berlin 2014, PyData Berlin 2015, PyData Berlin 2016, PyData Berlin 2017, PyData Berlin 2018]
| title | author | length |
|---|---|---|
| Keynote: Hacking the Iron Curtain: From smuggling computer parts to owning... | Andrada Fiscutean | 59m |
Students build Cobra computer clones around a Z80 CPU, soldering motherboards from factory rejects, custom video cards and cassette storage; fewer than a… | ||
| Q&A with Keynote – Andrada Fiscutean | Andrada Fiscutean | 14m |
Description of early 90s Romanian hacking culture with concrete artifacts and outcomes: hardware clones made by Electronica Factory in Bucharest sold as gaming… | ||
| Keynote: Building in Privacy and Data Protection -- what is demanded by the GDPR? | Marit Hansen | 59m |
Overview of GDPR plus concrete obligations and gaps: Data Protection by Design and Default is legally required but described in one abstract paragraph, so… | ||
| Q&A with Keynote – Marit Hansen | Marit Hansen | 29m |
Answers focus on concrete rules and fixes: GDPR allows processing under Article 6 with mitigations like encryption or pseudonymization, but consent blocks… | ||
| Keynote - Fairness and Diversity in Online Social Systems | Elisa Celis | 54m |
Algorithms systematically amplify real-world biases, shown by COMPAS recidivism, search-order effects on voting, skewed Google Images and biased ad delivery… | ||
| Q&A with | Elisa Celis | 41m |
Concrete claims about methods and deployments: a published algorithm gives fair classification with approximation guarantees across arbitrary group partitions… | ||
| Tricks, tips and topics in Text Analysis | Bhargav Srinivasa Desikan | 01h 28m |
Pipeline shows concrete choices: spaCy for POS, NER and lemmatization, gensim LDA on a 300‑article news corpus to surface ~5–10 rough topics, scikit‑learn +… | ||
| Semi-supervised learning with GANs | Andreas Merentitis, Carmine Paolino, Vaibhav Singh | 01h 27m |
PyData Berlin 2018
In many practical machine learning classification applications, the training data for one or all of the classes may be limited. We will… | ||
| Leveling Up Your Storytelling and Visualization Skills | Gerrit Gruben | 01h 27m |
Looks for a concrete claim (a single thing the work actually proves), a live demo of Jupyter notebooks and matplotlib/plotly or specific tooling rather than… | ||
| Deep Neural Networks with PyTorch | Stefan Otte | 01h 26m |
Hands-on walkthrough of PyTorch that implements networks from scratch and classic ML models with runnable notebooks and a dogs-vs-cats example showing training… | ||
| A Hands-On Introduction to Your First Data Science Project | Em Grasmeder, Jin Yang | 01h 23m |
A hands-on playground for demand forecasting: a runnable GitHub repo and Jupyter notebooks that use real Ecuador supermarket data (about 4,000 items, 50… | ||
| Scaling and reproducing deep learning on Kubernetes with Polyaxon | Mourad Mourafiq | 01h 18m |
Polyaxon runs on Kubernetes using Helm charts and a CLI to build docker images schedule jobs and track experiments with immutable artifacts and outputs It… | ||
| Search Relevance: A/B testing to Reinforcement Learning | Arnab Dutta | 57m |
Compares A/B testing with reinforcement learning for e‑commerce search relevance and claims multi‑armed bandits give faster convergence than A/B when quick… | ||
| Production-ready Data Science with Python and Luigi | Mark Keinhorster | 01h 27m |
Concrete production pipelines built with Spotify Luigi are shown, using runnable Python Tasks implementing parametrize and output to wire download → train →… | ||
| Deprecating the state machine: building conversational AI with the Rasa stack | Justina Petraitytė | 01h 27m |
Live code shows a concrete pipeline: Rasa NLU & Rasa Core plus SpaCy vectors, SVM intent classifier and conditional random fields for entity extraction, a tiny… | ||
| Deploying a Machine Learning Model to the Cloud Using AWS Lambda | Dr. Benjamin Weigel | 01h 27m |
Shows how a saved Jupyter model becomes a live HTTP endpoint: uploads model to S3, deploys Lambda + API gateway, and wires CloudFormation or the Serverless… | ||
| Visual concept learning from few images | Vaibhav Singh | 40m |
Describes concrete methods: using GAN’s and siamese networks to augment data and verify image pairs, episodic training and meta-learner LSTM that proposes… | ||
| Five things I learned from turning research papers into industry prototypes | Ellen König | 29m |
A practical five-step workflow converts papers into working prototypes: five-step workflow, search and evaluate by concrete metrics and baselines, prioritize… | ||
| Simple diagrams of convoluted neural networks | Piotr Migdał | 32m |
PyData Berlin 2018
Neural networks are complicated, multidimensional, nonlinear array operations. How can we present them in a way that is clear, didactic and… | ||
| Where NLP and psychology meet | Alexandra Klochko | 33m |
Claims that organizational culture can be measured continuously from collaboration text using NLP rather than surveys, and demonstrates a scoring engine built… | ||
| Spatial Data Analysis With Python | Dillon R Gardner, PhD | 37m |
Claims that lightweight Python tooling turns spatial data into actionable signals: a reproducible Jupyter workflow loads DHS cluster points, reprojects and… | ||
| Towards automating machine learning | Thorben Jensen | 36m |
Benchmarks run on Iris and a real insurance dataset with XGBoost on 16‑core machines over many random seeds show automated hyperparameter tuning routinely… | ||
| Deep Neural Networks for Double Dummy at Bridge | Lorand Dali | 40m |
A deep neural network predicts double dummy trick counts using TensorFlow with convolutional layers and ResNet-style skip connections and trains on millions of… | ||
| ctparse: a practical parser for natural language time expressions... | Sebastian Mika | 43m |
ctparse is a pure python MIT-License library that turns messy time phrases into structured time or interval objects by combining regex terminals with a… | ||
| Python Unittesting for Ethereum Smart Contracts or how not to create your own Crypto | Robert Meyer | 41m |
Claims that Python can unit test Ethereum smart contracts get shown with concrete tools and failures: Populus is used to create an in‑memory chain and accounts… | ||
| Launch Jupyter to the Cloud: an example of using Docker and Terraform | Cheuk Ting Ho | 29m |
Deploys a Jupyter notebook on GCP by building a Docker image (swapping to GPU variants), writing a startup shell that installs drivers and Docker, and… | ||
| Let's SQL Like It's 1992! | James Powell | 40m |
Models a turn based space game in SQL and shows concrete results using Postgres with a user plugin that implements full temporality so updates and deletes… | ||
| Building new NLP solutions with spaCy and Prodigy | Matthew Honnibal | 40m |
Claim: NLP projects fail when teams design pipelines and annotation schemes upfront instead of iterating. Built and used spaCy and Prodigy with Thinc-backed… | ||
| Python in Medicine: analysing data from mechanical... | Gusztav Belteki | 36m |
PyData Berlin 2018
I would like to show how I have been using Python and its data science tools to interpret large datasets retrieved from modern equipment… | ||
| Manifold Learning and Dimensionality Reduction for Data Visualization | Stefan Kühn | 38m |
Compares concrete manifold-learning and dimensionality-reduction algorithms (PCA, Isomap, Spectral Embedding, t-SNE) using scikit-learn implementations and… | ||
| A/B testing at Zalando: concepts and tools | Shan Huang, Grigory Bordyugov | 37m |
Overview of what actually happened and whether it’s useful tonight: Zalando describes real A/B infra with three services for bookkeeping assignment and… | ||
| How I Made My Computer Write it's First Short Story | Alexander CS Hendorf | 40m |
Uses PyTorch and a local eGPU / GTX 1080 to train models that attempt full episodes: character-aware text generation from fan transcripts, TTS trained on the… | ||
| How to Scare a Fish (School) | Andrej Warkentin | 23m |
Computational model connects a single-tap startle to neural response and emergent school behavior and claims a plausible mechanism for how local starts… | ||
| Extracting Relevant Metrics with Spectral Clustering | Evelyn Trautmann | 36m |
Spectral clustering groups hundreds of time-series metrics into compact sets so monitoring reduces to a few representatives rather than hundreds of noisy… | ||
| Solving very simple substitution ciphers algorithmically | Stephen Enright Ward | 41m |
Claims an algorithm that recovers monoalphabetic substitution keys when the language is known and spaces are preserved, encoding text as a bigram frequency… | ||
| Understanding and Applying Self-Attention for NLP | Ivan Bilan | 38m |
Claims that Transformer-style self-attention encoder replaces LSTMs for relation extraction and yields a new state of the art: experiments use Python with… | ||
| AI in Healthcare | David Higgins | 36m |
Claims that convolutional neural networks match clinicians on chest radiographs and in one report beat 15 doctors on tumour diagnosis while running 2–3× faster… | ||
| On Laplacian Eigenmaps for Dimensionality Reduction | Juan Orduz | 38m |
Claim: eigenfunctions of the Laplace-Beltrami operator give a low‑dimensional embedding; algorithm: build a similarity graph with Gaussian or k‑NN weights… | ||
| Industrial ML – Overview of the Technologies Available for Machine Learning | Alejandro Saucedo | 44m |
A survey of distributed, horizontally scalable Python machine-learning pipelines, covering motivations, technologies, best practices and caveats. It focuses on… | ||
| GDPR in practise - Developing models with transparency and privacy in mind | Łukasz Mokrzycki | 40m |
Concrete steps for making GDPR compliance practical include anonymization at collection, automated data pipelines with CI to enable selective deletion and… | ||
| When to go deep in Computer Vision... and how | Irina Vidal Migallón | 40m |
Decide by specifics: a quick OpenCV plugin for Unity baseline often runs on phone/hololens and uses cascade detectors for face/hand detection; measure against… | ||
| How mobile.de brings Data Science to Production... | Florian Wilhelm, Markus Schüler | 42m |
Personalization at mobile.de uses Hadoop/HDFS, PySpark and TensorFlow to build production systems: a logistic‑regression buyer‑intent model reaches 70%… | ||
| Privacy-preserving Data Sharing | Omar Ali Fdal | 39m |
An overview that gives concrete failures and workable tools: shows real reidentification examples (Netflix, Strava, Massachusetts health data), demonstrates… | ||
| Object detection to Instance segmentation: Learn to apply several algorithms... | Sujatha Subramanian | 36m |
Compares Faster R‑CNN, SSD, YOLO and an open‑source Mask R-CNN pipeline that uses ResNet‑101/feature pyramid backbones, an RPN, and a mask branch trained with… | ||
| Simplifying Training Deep & Serving Learning Models...using Tensorflow | Holden Karau | 37m |
Hands-on pipeline work: building feature prep that computes constants with TF Transform, packaging Python into Docker via Beam’s portability layer and running… | ||
| pyGAM: balancing interpretability and predictive power using... | Dani Servén Marín | 32m |
pyGAM is a new pure‑Python implementation of pyGAM offering splines, smoothing penalties, per‑feature lambda tuning via generalized cross validation… | ||
| The Face of Nanomaterials: Insightful Classification Using Deep Learning | Angelo Ziletti | 44m |
PyData Berlin 2018
Artificial intelligence is emerging as a new paradigm in materials science. This talk describes how physical intuition and (insightful)… | ||
| Going Full Stack with Data Science: Using Technical Readiness... | Emily Gorcenski | 30m |
Argues for using NASA’s Technical Readiness Level (TRL) scale to move data work from notebooks into production, with concrete checkpoints: TRL-like milestones… | ||
| Career Panel | Nakeema Stefflbauer | 01h 14m |
Panelists describe concrete career moves: internships turning into junior data warehouse roles, a chatbot and NLP projects (sentiment analysis, topic… | ||
| Interfacing R and Python | Andrew Collier | 32m |
Interfacing R and Python becomes practical: reticulate (available on CRAN and from GitHub) claims direct import of Python modules, running Python strings/files… | ||
| All that likelihood with PyMC3 | Junpeng Lao | 41m |
Shows how to extract actual likelihoods from a PyMC3 model: extract component logp and evaluate model_logp sum at test points, visualize likelihood surfaces… | ||
| Data versioning in machine learning projects | Dmitry Petrov | 35m |
Open-source DVC claims to make experiments reproducible by pairing code with the exact data version and pipeline commands rather than just restating the topic… | ||
| Extending Pandas using Apache Arrow and Numba | Uwe L. Korn | 33m |
Claim: Pandas gains drop-in custom dtypes by storing data in Apache Arrow buffers and compiling element logic with Numba, yielding a native string… | ||
| Big Data Systems Performance: The Little Shop of Horrors | Jens Dittrich | 39m |
Clear, measured comparisons show which engines win where: MapReduce vs in‑memory engines benchmarks expose massive slowdown on disk‑bound joins and show that… | ||
| CatBoost: Fast Open-Source Gradient Boosting Library For GPU | Vasily Ershov | 40m |
PyData Berlin 2018
CatBoost (http://catboost.yandex) is a new open-source gradient boosting library, that outperforms existing publicly available… | ||
| Meaningful histogramming with Physt | Jan Pipek | 34m |
PyData Berlin 2018
Histogram is a very simple and powerful statistical tool (disclaimer: it has its weaknesses too). Standard Python scientific libraries… | ||
| LightFields.jl: Fast 3D image reconstruction for VR applications | Hector Andrade Loarca | 42m |
Claims a novel sub-Nyquist reconstruction that recovers depth maps from a limited linear sequence of views, not just more examples. Method uses shearlet… | ||
| Practical examples of interactive visualizations in JupyterLab... | Jeremy Tuloup | 24m |
Shows how a custom JupyterLab extension uses Pixi.js and Jupyter Widgets to draw low‑level shapes, sync Python state to a WebGL/canvas front end, and animate… | ||
| nothing matches | ||