Schedule

Detailed information about the activities

Statistics for data science (Dan Nicolae)

• A data science pipeline
• Data exploration
• Statistical inference with resampling methods

Data science with Python (hands-on) (Dan Nicolae, Razvan Bunescu)

• Intro to Python
• Pandas and data frames
• Probability and simulations

Statistical learning (hands-on) (Dan Nicolae)

• Regression models and inference
• Prediction and classification

Intro to machine learning (Razvan Bunescu & ChatGPT)

• Feature vector representations
• ML for Classification
• ML for Regression
• Clustering

Machine learning (hands-on) (Razvan Bunescu & ChatGPT)

• ML algorithms in Python
  • Implementation using NumPy
  • The sklearn library
  • Visualization using Matplotlib
• Experimental evaluation of ML models
  • Linear vs. non-linear classification

Deep learning with neural networks (incl. hands-on) (Pawel Gasiorowski)

• Deep learning with Artificial Neural Networks
• Image Processing, Object Classification and Detection with Convolutional Neural Networks
• Implementation in Tensorflow Keras
  • Regression and gradient descent
  • Activation Functions, Feedforward Process, Error Functions, Optimizers, Backpropagation
  • Logistic regression and NNs for non-linear classification
  • Transfer Learning technique

Time-series analytics (Jože Rožanec)

• Time series analytics techniques: filtering methods, interpolation, extrapolation, prediction with ML
• Time series databases
• Python libraries for working with time-series data

Introduction to graph data (incl. hands-on) (Dumitru Roman, Brian Elvesæter, Radu Prodan)

• Introduction to graph data
• Knowledge Graphs
• NoSQL databases
• Graph databases (focus on Neo4j)
  • Data model and data modeling
  • Query language
  • Graph algorithms / analytics / ML
• Introduction to massive graphs

Knowledge graphs for conversational AI (Ioan Toma)

• Introduction to Semantic Knowledge Graphs and their role in building intelligent chatbots
• Understanding knowledge modeling and ontology development for building Knowledge Graphs for Conversational AI
• Data import and mapping techniques to populate the Knowledge Graphs
• Overview of conversational setup and designing a chatbot interface
• Building a chatbot using Onlim Conversational AI framework
• Integration of Knowledge Graph data with the chatbot using API calls
• Querying and accessing Knowledge Graph data through Chatbots

FAIR data and best practices in data sharing (Anna Fensel)

• Introduction to FAIR data
• How to make data FAIR?
• Sharing data effectively with semantic technology:
  • Open data vs. closed data
  • Consent, contracts, licenses, legal compliance
  • Research data infrastructures

Data enrichment (Dumitru Roman, Nikolay Nikolov)

• Data preparation; cleaning, annotating and enriching data
• Semantic data enrichment
• Tools for semantic enrichment
• Data enrichment pipelines
• Example application for data enrichment

Operationalizing machine learning pipelines (Wiktor Sowinski-Mydlarz)

• What are Machine Learning pipelines
• Introduction to Software Containers and Cloud
• Deployment, orchestration, monitoring of ML pipelines on the Cloud – using python libraries
• Applications example

Required software

Software tools/services to be used during the sessions and hands-on include:

• Anaconda (https://www.anaconda.com): Installation instructions for various platforms can be found at: https://docs.anaconda.com/anaconda/install
  • A number of relevant tools and libraries that we will use can be configured from Anaconda: Python 3, NumPy, SciPy, Matplotlib, Jupyter Notebook, Ipython, Pandas, and Scikit-learn.
  • For Deep Learning we will use Keras Tensorflow (https://keras.io) to build and train ANN/CNN models. This can be installed from Anaconda, with ‘conda’ from the command line, the actual command line depends on the platform and version (conda install -c conda-forge keras)
• Onlim Platform (https://app.onlim.com/): Conversational and Knowledge Graph Platform. Accounts can be created https://auth.onlim.com/auth/realms/onlim/login-actions/registration?client_id=onlim&tab_id=gmTCMEh3-6U
• DataGraft (https://datagraft.io): Software as a service that requires sign up for a free (DataGraft platform account at https://datagraft.io/users/sign_up where you will be required to provide a username, email and password to access the website).
• Neo4j (https://neo4j.com): Installation and documentation can be found at https://neo4j.com/developer/get-started.We will use the online sandbox service provided at  https://neo4j.com/sandbox, so no installation on local machines is needed for experimenting with Neo4j. Alternatively you can download and install Neo4j Desktop, which provides a convenient way for developers to work with local Neo4j databases (this can be downloaded from https://neo4j.com/download-center/#desktop). We will also use Neo4j Graph Data Science (https://neo4j.com/product/graph-data-science) which comes with Neo4j.
• Docker (https://www.docker.com): An open-source containerization platform that will be used for ML pipelines. Installation instructions can be found at https://docs.docker.com/engine/install.