Bio

Arlette has a background in Artificial Intelligence which she studied at Utrecht University. She specialised in Agent Systems and Natural Language Processing, and researched human-computer collaboration at Harvard University. During her PhD Arlette published work on decision support for behavior change, persuasive technology, and computational models of (social) behavior and habits. In 2010 she was awarded the Google Anita Borg Memorial Scholarship for Women in Computer Science. Arlette currently works at Philips Research as a Senior Scientist, focusing on human-centered AI. She leverages AI technologies such as conversational intelligence and content personalization to increase patient and user engagement. At Philips, Arlette is driving the work on responsible AI innovation, finding solutions to challenges with respect to the ethical and societal impact of AI.

Abstract

For AI to be designed, developed and used responsibly means to create AI that is trustworthy, fair, robust, and transparent. In particular in a healthcare context, the growing use and adoption of AI can uncover AI’s vulnerabilities. Responsible AI development and deployment is therefore of growing interest to healthcare providers, patients, and regulators. This talk will cover why Responsible AI is of crucial importance in the healthcare domain, and how it can enable us to develop reliable and effective solutions to meaningful problems. In particular, it will demonstrate the work we do in Philips on creating Responsible AI tooling and identifying best practices, and how these are integrated into our workflows in order to create high-quality and trustworthy AI solutions. 

Bio

Yolanda van Dinther is Director Software Development in the Electron Microscopy division of Thermo Fisher Scientific, leading Digital Transformation with a focus on unlocking and using the power of big and large data through AI technology. 

Also, she leads the local chapter in Eindhoven of the international women’s network in Thermo Fisher Scientific, guiding and advancing female talent in the company.  

Before joining Thermo Fisher Scientific 4 years ago, Yolanda worked in various innovation leadership roles, ranging from software to system development across a variety of high-tech companies in the healthcare, semiconductors, consumer, automotive, life sciences and printing industries. 

Yolanda holds a MSc degree in Computer Science from the University of Technology in Eindhoven and a Post Bachelor’s degree in Counseling Psychology. 

Abstract

Thermo Fisher Scientific is the world leader in serving science. 

Through our Electron microscopy solutions our customers can combine high-resolution imaging with physical, elemental and chemical analysis to go from questions to usable information. 

In this talk we will demonstrate examples of data science applied to our imaging and system data. 

Bio

Duygu Büyükaydin received her MSc in Signal Processing in 2006 from the Electrical&Electronics Engineering department of Middle East Technical University, Turkey. Duygu worked in several projects in defense domain and gained experience in signal processing and software development. She worked as a Data Scientist in Philips Healthcare. Currently she is Lead Engineer in Thermo Fisher Scientific. She is responsible for data science track of digital innovation. ​

Abstract

Thermo Fisher Scientific is the world leader in serving science. 

Through our Electron microscopy solutions our customers can combine high-resolution imaging with physical, elemental and chemical analysis to go from questions to usable information. 

In this talk we will demonstrate examples of data science applied to our imaging and system data. 

Bio 

Miranda Roosen is Head of Data Business Enablement, an internal PostNL department that consists of the teams Analytics & Decision Support and Machine Learning Operations.  

PostNL is transforming into an insight-driven organization, in which insight from data is essential. Data Business Enablement is an organizational unit with a clear focus on producing data-driven insights and decision support by realizing the industrialization, availability, continuity and performance of (machine learning) algorithms that provide insights for PostNL. These data science products run in the core processes of PostNL and therefore have a strong operational responsibility. 

Abstract 

PostNL delivers parcels every day and mail five days a week and is the indispensable link for customers between senders and recipients. PostNL is also the connector between the physical and the digital world. PostNL is increasingly developing into a technology-driven organization in which data and analytics play a crucial role. A fully connected data ecosystem is needed to enable PostNL’s tech-driven future strategy. The data ecosystem is created by the Insights department, which makes the department a prerequisite for a successful digital transformation. Data Business Enablement is part of PostNL Insights. 

In this talk we will demonstrate examples of Data Analytics at PostNL. 

Bio 

Emmy Manders currently works as a research manager and data scientist at CytoCypher, a small biotech company that aims to improve data quality and increase throughput in muscle research by developing laboratory equipment and intelligent software solutions. She holds a master in Biomedical Engineering from Eindhoven University of Technology and a PhD in Physiology from VU University Medical Center. Her work focusses on streamlining the complete process of automating data acquisition, data analysis and visualization in cells derived from the heart. Additionally, she holds a part-time post-Doc position at the Amsterdam Medical Center assisting in large drug screening projects to unravel new targets for heart failure. 

Abstract: In fundamental biomedical research low numbers (10-200) of data were and still are quite common and the subsequent data management, analysis, visualization can be handled with simple tools such as Excel or GraphPad.  

With image recognition and inventive hardware, we’ve been able to automate data acquisition and increased the throughput. With the increasing number (10-30K) of data new challenges and opportunities arise for biomedical researchers analyzing, managing and interpretating the data. 

Bio

Sofía Calero studied Physical Chemistry at University Complutense of Madrid, where she received her MSc degree in 1995. In 2000, she obtained her PhD, cum laude and extraordinary prize, at the same university. From 2001 to 2003 she was a postdoctoral researcher as a Marie Curie Fellow in the Chemical Engineering Department of the University of Amsterdam, the Netherlands. In 2004 she moved to the University Pablo de Olavide, Seville, Spain as Ramon y Cajal Fellow (2004), Profesor Contratado Doctor (2006), Profesor Titular de Universidad (2009) and Catedrático de Universidad (2017). In 2020 she was appointed full Professor and chair Materials Simulation & Modelling at the department of Applied Physics, at Eindhoven University of Technology (the Netherlands).  Calero received several grants and awards, including the Marie Curie Excellence Award (2005), ERC Consolidator Grant (2012-2016), Salvador de Madariaga Grant (2016), Dutch VPP-KNAW grant (2017), Spanish Royal Society of Chemistry awards for Young Researchers (2005) and for Scientific Excellence (2018) and Irene Curie Grant (2020).  Her research involves the application of molecular simulation to industrially relevant systems and the development of force fields, algorithms and simulation methods to reverse-engineer properties of porous materials.  

Abstract

We could imagine an economy in which renewable energy is used in combination with materials to cost-effectively improve industrial processes. However, this is still projected into the future, as it requires stable and synthesizable materials with optimal performance. I propose to develop a new framework inspired by the concept of information contained in deoxyribonucleic acid (DNA). Just as each species has a differential DNA, my approach is to build a kind of in silico DNA for materials. In the same way that the DNA of humans marks their traits, this DNA will contain all the information related to the properties of the material. If biological DNA is made up of base pairs, the DNA of these materials will be made up of a series of descriptors that define their properties and performance. Unlike previous methodology using conventional “brute force” machine learning, the framework we are suggesting will use a multi-layer machine learning architecture. The challenge will be in identifying, generating, and assigning the descriptors, relating the descriptors to measurable observables, and filling in the DNA gaps to get complete sequences. The final step would be to apply inverse design to predict new stable, high-performance materials with increasing efficiency and speed. In this way, we will benefit from the advantages of using in silico models, but at the same time we will bridge the gap that now exists between design and applicability, thus changing current concepts and opening a new avenue for materials discovery 

Links 

www.upo.es/raspa 

www.tue.nl/msm 

Bio

Dr. Ana Karla Alves de Medeiros is a Brazilian with a background in Computer Science. About 20 years ago she moved to the Netherlands to pursue her PhD in the area of process mining at the Technical University of Eindhoven (TU/e). Since then, Ana Karla has developed and applied her knowledge and passion to empower businesses with actionable data-driven customer experience insights and solutions. And this is exactly what she is doing in her role as a Senior Data Scientist at Jumbo! 

Abstract

Jumbo has become an omnichannel supermarket in the last years. This talk illustrates some of our data science applications to help Jumbo keep delivering a good customer experience both on bricks and clicks. More specifically, we will show a solution to perform voice of the customer mining at our service centre, so that continuous improvement in our value chains are driven by customer needs. The talk will conclude with some personal notes on my journey as a woman in data science. 

Bio

I finished my masters in Biomedical Engineering at the University of Twente with the specialisation in Robotics. 

At Demcon I have worked as a Systems Engineer in the field of Medical and High Tech for four years. I then joined the robotics department when it was founded 2 years ago and am now leading different robotics and machine learning projects that include anomaly detection, simulation and reinforcement learning.  

Abstract

In computer vision and machine learning in general, the current limitations the industry faces today is the availability of large, varied and annotated data sets for training. In many cases, data gathering and labeling is extremely time consuming and expensive or sufficient unique images are simply not available. Creating synthetic data provides the solution to these problems and has many other advantages. For anomaly detection of manufacture products, synthetic data allows for inclusion of rare edge-cases of unpredictable deformaties in datasets. Complex annotation with 100% accuracy can be achieved. In robotics, a simulated 3D environment can be sampled as input for digital sensors of a  digital model of the robot, required to apply reinforment learning. Additionally, procedural generation of data allows for near infinite permutation of the environment to ensure a varied and rich data that includes a perfect ground truth for the robot model.  

We will look at these and other use cases for synthetic data generation, how unique photorealistic images can be created procedurally using techniques from the field of digital 3D art and animation, and how the gap between synthetic data sets and real data sets can be decreased.  

Bio

Véronique Van Vlasselaer is the Analytics and AI lead for South, West and East Europe at SAS, and a true data science enthusiast. In her job, she passionately helps companies to envision and prepare for an AI-driven future, embrace the power of data science to support intelligent decisioning, and discover the real value in their data. Before she joined SAS, she graduated as Doctor in Business Economics at the KU Leuven (Belgium) with the department of Information Management and Decision Sciences. She is co-author of Fraud Analytics Using Descriptive, Predictive, and Social Network Techniques. 

Abstract

Artificial Intelligence is said to be one of the most important drivers in the future of our society, automatically analyzing massive amounts of data, distilling new knowledge, and making automated intelligent decisions, beneficial to all. Without realizing, AI systems facilitate, influence and augment many aspects of our life. But are these AI systems truly intelligent and beneficial to all? More and more incidents of discrimination by AI surface, putting AI in a bad light.  

In this talk, Véronique Van Vlasselaer questions the criticisms on AI neutrality and claims that before pointing the finger to AI, we should first look at ourselves and our society: the explicit bias in AI models and systems is merely a translation of the often unspoken bias in our society. Only through a good collaboration between humans and AI we will be able to make our world a little better. And a world of equality and without prejudice – on every level – is what we are striving for. 

Bio

Ymke is working as a data science professional in the Philips global data&AI Center of Excellence at Royal Philips with a current focus on data sharing and data&AI partnerships. Her mission is to connect the academic cutting edge technologies to the practical challenges in healthcare we are facing right now. Seeing data driven solutions actually have impact and meet and exceed expectations gives her energy. Ymke has a background as biomedical engineer and has experience in clinical, insurance and data architecture field. 

Abstract

Healthcare providers face an unsustainable burden, while patient expectations of healthcare are changing. In the same time, the amount of healthcare data is growing exponentially and artificial intelligence can help us providing insights out of this data and help solve the current challenges in healthcare. However, we do see limited AI solution being implemented and adopted in healthcare. In this talk we will discuss the challenges of scaling data&AI solutions in healthcare and how we can work together to solve those challenges.  

Bio

Blagorodna is a Data Scientist with a background in Computer Science and Electrical Engineering.  

Before starting her education in Data Science, she worked for 5 years as a Software developer for various applications in the Health and Cyber security domain.  

After completing her Professional Doctorate in Data Science at the Technical University of Eindhoven (TU/e), , she started working as Data Scientist in ASML. In this role, she has been afforded the opportunity to work on several industry related projects that cover the full scope of Data Science starting from Data Visualization to Machine Learning and AI. 

Abstract

Wafer defects are found during qualifications tests and can indicate damages or particles presence on the wafer. The cost of manually identifying such defects is prohibitive.  

A Machine Learning (ML) algorithm could be used to automate this task and provide consistent performance. However, the lack of large accurately labeled datasets and the extreme class imbalance challenge the accuracy of ML algorithms.  

To deal with these challenges, we integrate domain expertise into ML model to improve its performance using Active Learning (AL). An AL system improves the ML algorithm by proactively selecting only the most informative and uncertainly classified images for human input from a pool of unlabeled data. The goal of this process is to query the most informative examples so that the model achieves the best possible performance given the least amount of labeled training data.  

In this talk, we show the improvement of the performance of a deep Convolutional Neural Network Classifier using Active Learning.

Bio

Renata Medeiros de Carvalho is an Assistant Professor (Universitair Docent) at Eindhoven University of Technology (TU/e) since 2016. She received the B.Sc. and M.Sc. degrees in computer engineering from the University of Pernambuco (UPE), Brazil, in 2007 and 2010, respectively, and the Ph.D. degree in computer science from the Center of Informatics (CIn), Federal University of Pernambuco (UFPE), Brazil, in 2015. She was a Postdoctoral Researcher with the Laboratory for Research on Technology for Ecommerce (LATECE), University of Quebec at Montreal (UQAM), for more than one year. Her research interests are focused on business process management, discover and improvement. She focuses the work on the domain of flexible/dynamic/adaptable business processes, and how it can enrich business process models with domain-specific knowledge. Currently, her research is focused mainly on the domain of healthcare. 

Abstract

With an aging population and shortage of care staff, hospitals are faced with the urge for new sustainable care models that are effective while ensuring the care needs of patients are met. In this talk, I will present how Process Mining has been applied to (re-)design sustainable health care processes. On the one hand, the technical challenge is to deal with huge amounts of patient data that follow different care paths, to be able to learn important and effective polices, and to define how they can be re-used in a more sustainable way. On the other hand, there is the social challenge to deliver top quality care while balancing costs (and resources used) and benefits. 

In case you want it:  

• https://www.win.tue.nl/~rmedeiro/