The Chief Architect & Data Management department CADM has outlined 7 different assignments that you can apply for. Please choose only ONE of the assignments to apply for.
1. Deep learning approach to automated receipt understanding. Input: Public datasets of scan receipt. Desired output: Extract information from receipts such as, items, quantity, price, total price, discount per item, date and time of receipt, name and type of receipt. Requirements: Python, Pytorch, OCR, Image processing, NLP.
2. Cross-Organisation Federal Learning. Input: Fake dataset for experimentation purposes. Desired Output: A working POC with working server (ABN AMRO) and client (Third party, can be fake), where the client is able to train on the data of the server without seeing the real data of the server. Requirements: Python programming experience, Deep Learning, Machine Learning.
3. Measuring the performance of XAI approaches. Input: one (or potentially two) use cases will be identified to analyse the ‘goodness’ of explainability and how appropriate it is for different key stakeholders of the model. Desired Output: evaluation metric/scheme of (a few promising) explanation approaches. Requirement: python (desired), or R (alternative).
4. Smart Web Scraping: An intelligent method for content scarping. Input: Sources, current work on this subject. Desired Output: Script/library that can automatically scrape and link of an article. Requirement: Machine Learning, Python, Information retrieval, Web scraping
5. Text categorisation for Content Filtering. Input: Scraped articles from several sources. Desired Output: Model of or report on the feasibility of classifying the articles into this ontology. Requirements: Machine Learning, Python, NLP (Recommended), Dutch Language.
6. Text Analysis on Real Estate valuation reports. Input: Dataset with many fields, among which are text fields from valuation reports. This dataset contains the valuation reports thousands of houses. Desired Output: Insights into and prediction of house prices in relation to these descriptions. Requirement: Python, NLP, Machine Learning, Deep Learning (Recommended).
7. Text Analysis for real estate market forecasting. Input: National and local news, real estate websites, CBS data, economic financial publications. Desired Output: Forecasting model for the current moment, as well as say 3-6 months in the future. Requirement: Python, Webscraping (optional), NLP.
Stage4eu is free of charge and has no commercial purpose. It does not conduct brokerage activities, nor does collect CVs. By clicking on the green button “Vai all’Offerta” you’ll be redirected to the original vacancy posted on host organizations’ web page.