Women As Research Peers for Information Technology
Translated with DeepL.com
You are a student of mathematics, computer science, natural sciences or technology (MINT) at KIT and wonder what comes after your studies? You would like to learn why your knowledge is particularly in demand in computer science and data science? You would like to see a supercomputer from inside and want to know what female scientists in computer science research all day long?
Then apply for an exciting research project in the mentoring program.
Fact is: women are still underrepresented in computer science. We, an association of female scientists at the Steinbuch Centre for Computing (SCC), want to change that. We have set ourselves the goal of giving female STEM students in particular an insight into our working lives as scientists. In this context, we offer:
- Three positions as an assistant scientist for six months, each on an interesting project from current research in computer science and data-driven computing (see below for project descriptions).
- Supervision by experienced and dedicated SCC researchers and staff as a personal contact around the job as an assistant scientist and IT-ler (see below)
- Contact to like-minded people and concrete opportunities for networking and exchange
- Attractive training and further education opportunities as well as subject-specific information on a possible career in IT
Apply with your CV and tell us in a short motivation letter (half page) which of the three projects suits you best and why. German or English welcome!
Application until April 22, 2022 to warp4it∂lists.kit.edu.
All projects start on 01.05.2022 and run through the summer semester 2022.
- Commitment and motivation to work on your own project (conception and implementation).
- Interest in interdisciplinary collaboration
- Affinity for technical concepts and methods
- Enthusiasm for computer science and data science research
- Basic knowledge of programming is helpful, but not mandatory.
A. How to make our power supply more reliable?
Artificial intelligence(AI) plays a crucial role in the analysis and prediction of time series. One example is the application of AI for controlling and optimizing electric power grids - a challenge that is becoming ever important in light of the upcoming energy revolution. Using modern AI algorithms, we are able to predict the energy consumption on many time scales, which can help to improve grid stability and thus secure the supply. However, the growing application of AI for such critical infrastructures naturally presents us with the following question: To what extent can we trust the forecasts of our models?
To tackle this challenge, researchers have developed and investigated a wide variety of methods for quantifying uncertainties in deep neural networks. In this project, we want to apply such methods to AI models for time series forecasting in order to evaluate their "trustworthiness" by estimating their uncertainties and error rates. For this purpose, you will implement a specific neural network called "Sequence2Sequence" model in Python and train it on electricity consumption data from different European countries. Afterwards, you will determine the prediction uncertainty as a reliability measure of our network using two different approaches, that is the "Prediction Intervals" (PIs) and the "Lower Upper Bound Estimation" (LUBE) method.
B. How to read a pyramid like a book?
Egyptian pyramids are ancient evidence of a longgone culture. Inside you will find a variety of hieroglyphs called "spells" that provide protection and good wishes for the afterlife of the deceased. Which spells were used for which context and how they are connected are questions that are of great interest to current researchers. Within this project, you will investigate the "International Image Interoperability Framework" (IIIF). IIIF is a framework to standardize the handling of images and audio/video files on the web to, e.g., allow reuse of research data and tools. Currently mostly used for manuscripts and books, we want to find out about potentials, pitfalls, and of course the applicability to pyramids to allow insights into this unique data treasure.
C. How to enhance CNN-based learning with quantum computers?
In the so-called noisy intermediate scale quantum (NISQ) era, devices have a limited number of qubits with little error correction. Thus, developing useful computational algorithms for NISQ devices is an urgent task at present. In this domain, the field of quantum machine learning (QML) is gaining attention in hope of achieving computational speedup or better performance for some machine learning tasks. Especially parameterized quantum algorithms provide a possibility for QML towards quantum advantage in the NISQ era. Compared to conventional quantum algorithms, parameterized quantum algorithms are inherently robust to noise and could benefit from quantum advantage by taking advantage of both the high dimensionality of quantum systems and the classical optimization scheme.
Within this project, you will design and analyze a hybrid quantum-classical algorithm by incorporating a parameterized quantum algorithm into a classical convolutional neural network (CNN). Our goal is not only to validate the classification performance based on the well-known MNIST dataset in comparison to a classical CNN but also to analyze potential benefits of noise robustness. The implementations of this project will be done in Python and requires no prior knowledge of quantum computing.
The percentage of women in computer science and other IT-related professions is low. For several years, female first-year students have accounted for only about a quarter of all computer science students nationwide. We, an association of female research assistants at the Steinbuch Centre for Computing, would like to change this, because we experience our professions in computer science and data science research as diverse and enriching. Digital processes and IT systems of all kinds have become an integral part of our society. However, the underlying research and development still takes place from a predominantly male perspective. That is why we are committed to implementing equal opportunities and equal participation of women in information technology. We want to show female STEM students their high potential for actively shaping our society and raise awareness of the great competence of female computer scientists in research and careers. To this end, networking, female role models and practical insights into everyday working life are particularly important (see also Women in Science & IT at SCC). Warp4IT creates exactly that: In addition to interesting projects from current research, which are worked on in the context of an employment as an assistant scientist, we offer concrete opportunities for networking, further education and exchange as well as subject-specific information for a career path in IT & Co.
... researches on artificial intelligence methods for the energy system as a Helmholtz AI consultant.
Artificial Intelligence (AI) has become a central part of our daily lives: from face-recognition algorithms for smartphones over voice-assistants like Alexa or Siri, to personalized movie recommendations on Netflix. But also scientific research has benefited immensely from the advances in AI in the past decade, leading to novel insights and innovation in many domains. As an example the development of future energy systems in order to implement the transition towards renewable energy sources and thus tackle climate change relies heavily on modern AI methods.
The goal of our Helmholtz AI team is to bring expertise on state-of-the-art AI methods to research groups in the field of “Energy” all across the Helmholtz Association. As such, we offer consulting with respect to application, implementation and validation for these algorithms, and help groups to access the necessary compute resources here at SCC.
Being an AI consultant offers new prospects and challenges every day, as the applications of AI in energy research are numerous and divers: Be it the prediction of electrical load or solar surface irradiation for photovoltaic installations, or the monitoring and control of production processes for solar cells, AI can assist in understanding and improving all of them.
Contact: Dr. Charlotte Debus
... is researching hybrid algorithms in the field of quantum machine learning.
Today's quantum computers belong to the so-called NISQ devices, since they have only a small number of qubits, which are also very error-prone. Nevertheless, the advantages over classical computers can already be investigated on this basis. In particular, the use of hybrid algorithms, where parameterizable circuits for quantum computers are trained by classical optimization methods, are promising.
My team and I are investigating such hybrid algorithms and are concerned not only with practical implementations in the area of quantum machine learning, but also with scalability and generalization for future devices and potential use cases.
I also actively advocate for the sustainable development of research software and its importance in science.
Contact: Dr. Eileen Kühn
...researches how humanities and cultural studies can be enriched by research data management methods and is deputy head of the DEM department.
In the Data Exploitation Methods (DEM) department, we research new methods for research data management and analysis. Together with expert scientists from various disciplines, we take an interdisciplinary look at research questions that were previously difficult to solve or even impossible to solve.
Especially in disciplines that have hardly worked digitally so far, for example the so-called 'small subjects' of the humanities and cultural studies, there is immense, still undecided potential of interdisciplinary research. Here we have the chance to enrich the methods and ways of working of the entire discipline.
As part of the Collaborative Research Center 980 'Episteme in Motion', in which about 50 humanities scholars are investigating the transformation of knowledge over several millennia, I am leading the information infrastructure project 'Books on the Move'. As a central task, we provide methods and tools - from sustainable storage of research data to annotation and visualization of specific phenomena - for the entire research network.
Contact: Dr. Danah Tonne
... works as a Helmholtz AI Consultant on artificial intelligence methods for energy research.
After graduation in physics from KIT and completing my PhD in computational biophysics at SCC, I switched to computer science as a research associate. Together with my colleauges, I work together with my colleagues on artificial intelligence (AI) methods for the energy system of tomorrow. AI is a growing component in our everyday lives, taking on increasingly important tasks in our digital society, besides tailored music recommendations on Spotify and personalized shopping ads on Instagram & Co. The application areas in energy research are as diverse as the field itself and range from load forecasting for energy systems and the development of new materials for batteries to the automated control of industrial plants.
Helmholtz AI's consulting concept is designed to enable scientists in the Helmholtz Association to use cutting-edge AI methods for their own research. As Helmholtz AI Consultants, we provide support with our expertise in AI methods and software engineering for concrete research projects. This interdisciplinary work is very diverse and offers, in addition to exciting research questions, many opportunities to learn new things and to get in touch with researchers beyond disciplinary boundaries.
Contact: Dr. Marie Weiel