Background and summary of fellowship
In many application areas, it is not sufficient to present the output of machine learning models to the users without providing any information on what leads to the specific predictions or recommendations and how (un)certain they are. The strongest machine learning models are however often essentially black boxes. In order to enable trust in such models, techniques for explaining the predictions in the form of interpretable approximations are currently being investigated. Another cornerstone for enabling trust is that the uncertainty of the output of the machine learning models is properly quantified, e.g., that the output prediction intervals or probability distributions are well-calibrated.
Motivated by collaborations with Karolinska Institutet/University hospital on sepsis prediction, Scania on predictive maintenance and the Swedish National Financial Management Authority on gross domestic product (GDP) forecasting, techniques for quantifying uncertainty and explaining predictions will be developed and evaluated. In addition to scientific papers, the output of the project will be Python packages to support reliable machine learning, enabling predictions of state-of-the-art machine learning models to be complemented with explanations and uncertainty quantification.
Background and summary of fellowship
Social robots and virtual agents are currently being explored and developed for applications in a number of fields such as education, service, retail, health, elderly care, simulation and training and entertainment. For these systems to be accepted and successful, not only in task-based interaction but also to maintain user engagement, in the long run, it is important that they can exhibit varied and meaningful non-verbal behaviours, and also possess the ability to adapt to the interlocutor in different ways. Adaptivity in face-to-face interaction (sometimes called mimicry) has for example been shown to increase liking and affiliation.
This work addresses how style aspects in non-verbal interaction can be controlled, varied and adapted, across several modalities including speech, gesture and facial expression. The project entails novel data collection of verbal and non-verbal behaviours (audio, video, gaze tracking and motion capture) with rich style variation, but also makes use of existing datasets for base training. Synthesis models trained on this data will be based primarily on deep probabilistic generative modelling, conditioned with relevant style-related parameters. Multimodal generation paradigms, that produce congruent behaviours in more than one modality at a time, e.g. both speech and gesture, in a coherent style, will also be explored and evaluated in perceptual studies or experiments with real interactive contexts.
About the project
Objective
The objective of the Swedish Question Generation for Assessing Reading Comprehension (SWE-QUEST) project is to develop a demo system that, given a text, automatically generates multiple-choice reading comprehension questions on the text, as shown in the picture below.
Using a neural-network-based generative model, the demo system will generate the whole multiple-choice question, both the question itself and the answer alternatives. Although the example above was in English, our demo system will work for Swedish text. The project will push the state-of-the-art in natural language generation. The system is intended to be used by teachers of SFI (Svenska för invandrare) to facilitate test construction and the development of teaching materials. Still, it can also be used for self-studies of Swedish.
Background
The SFI students form a heterogeneous group, and the SFI classes are often too big to allow the teacher time to adapt the level of teaching to cater to each student individually. Therefore, developing such a tool could be highly useful. It would rapidly and easily generate several suggested multiple-choice questions (MCQ) on text material for teaching and assessment.
The task of automatically generating reading comprehension questions (without distractors) using neural methods has been studied before, primarily for English. There have also been some attempts at generating distractors, given the question and the correct answer, using neural methods. However, no attempt has been made to construct a trainable model that generates the whole MCQ in one go.
Crossdisciplinary collaboration
The researchers in the team represent the KTH Schools of Electrical Engineering and Computer Science, the Division of Speech, Music and Hearing and the Department of Swedish Language and Multilingualism at Stockholm University.
Watch the recorded presentation at the Digitalize in Stockholm 2023 event:
About the project
Objective
Susan’s ride on Campus2030 aims to demonstrate the potential of digitalization in reducing the carbon footprint and improving the cost-efficiency of the construction and transportation industry. With this objective, the project will establish a one-of-a-kind smart road infrastructure demonstrator on the KTH campus Valhallavägen for the integrated design, construction and operation of smart infrastructures. The demonstrator will incorporate a digital twin of the KTH campus, corresponding to multiple models and data sets that enable virtual assessment and experience of the Campus infrastructure while being validated and updated through real-time data feeds from various sensors. Our work in this direction can be seen on our testbed for Intelligent Transportation Systems on www.adeye.se. Susan’s ride will showcase the potential of edge computing, federated learning, and digital twins in the digital transformation of road construction and autonomous vehicle path planning.
Background
Autonomous vehicles, dynamic charging of electric vehicles and vehicle-to-infrastructure communication are just a few examples that require a systemic solution to function sustainably. Making the smart road sustainable requires a partnership between road owners, operators, electricity companies, vehicle manufacturers, transport and logistics companies, and technology suppliers in digitalization. Data will become a fundamental asset in this partnership. They must be collected through a combination of new sensors in the infrastructure already upon construction on smart vehicles, including construction machinery.
Crossdisciplinary collaboration
The researchers in the team represent the School of Electrical Engineering and Computer Science, KTH, the School of Architecture and the Built Environment, KTH and the School of Industrial Engineering and Management, KTH. The project leverages and extends research carried out in the Campus 2030 project and the TECoSA research centre.
Watch the recorded presentation at the Digitalize in Stockholm 2023 event:
About the project
Join the Second Drone Challenge at the Digital Futures Drone Arena, a one-of-a-kind interactive event with aerial drone technology! This year, the challenge focuses on moving with drones in beautiful, curious, and provocative ways – without needing to write a single line of code. The event takes place on May 16-17, 2023, at KTH’s Reactor Hall in Stockholm, Sweden. Read more about the challenge, the prizes, and how to sign up on our Drone Arena challenge website.
Objective
The Digital Futures Drone Arena is a concrete and conceptual platform where key players in digital transformation and society join in a conversation about the role and impact of mobile robotics, autonomous systems, machine learning, and human-computer interaction.
The platform is a novel aerial drone testbed, where drone competitions occur periodically to understand and explore the unfolding relationships between humans and drones. Aerial drones are used as an opportunity to create a foundation that lives past the end of this project. It is a long-standing basis for testing technical advances and studying, designing, and envisioning novel relationships between humans, robots, and their functioning principles.
Background
Few robot testbeds exist to experiment with application-level functionality. The Digital Futures Drone Arena bridges this gap by providing an easy-to-use programmable drone testbed for experimenting with novel drone applications and exploring the relations between humans and drones. The latter activity is driven by the concept of a ‘soma’ or the lived and felt body as it exists, moves, and senses the world. The theory provides an ethical stance on the soma, highlighting how technologies and interactions encourage certain movements and practices while discouraging others. As a critique of technology design and use, somaesthetics addresses the limited and limiting ways we sit at desks and tap away at keyboards. When we interact closely with drones, we must adapt to how we control them and move around them.
Crossdisciplinary collaboration
The researchers in the team represent the Connected Intelligence Unit, RISE and the Department of Computer and Systems Sciences, Stockholm University.
Articles:
Watch the recorded presentation at the Digitalize in Stockholm 2023 event:
About the project
Objective
The main aim of this project is to increase the diagnostic power of PET images for detecting lung cancer lesions at an early stage by overcoming the loss of contrast and spatial resolution caused by respiratory motion during data acquisition. Traditional ways of tackling this problem are computationally too demanding to be useful in clinical practice. For this reason, we will implement algorithms deploying a mixture of modelling of the data-acquisition set-up and machine-learning-based tools for image registration. The challenges of this project consist mainly in tackling the size of this four-dimensional image reconstruction problem and in being able to deliver images to radiologists in a time frame compatible with the hospital workflow.
In collaboration with our clinical partners at Karolinska Hospital in Huddinge, we will collect gated PET line-of-response activity data and corresponding 3D CT attenuation maps of the thorax region of 400 patients and test and optimise our 4D algorithms and gating strategies on this data. We will evaluate the resulting reconstructions with a particular focus on their diagnostic power for small lung lesions and make the reconstruction software package openly available.
Background
PET (Positron Emission Tomography) is a medical imaging modality that reconstructs the 3D distribution of metabolic activity by detecting photons emitted during the in vivo annihilation of free electrons with positrons from an injected radioactive tracer. In principle, cancer lesions will be visible in the reconstructed image with high contrast compared to the surrounding healthy tissue thanks to their peculiar metabolic fingerprints (e.g. higher sugar metabolism). PET is, in fact, one of the most powerful imaging modalities for cancer diagnosis and staging. However, the long acquisition time required to collect projection data with an acceptable noise level leads to motion artefacts that strongly affect the contrast-to-noise-ratio of lesions. This is of particular interest when trying to detect tumours that are of a size comparable to the system resolution (~5 mm) and that are continuously moving because of respiratory motion. Those lesions are the most important ones to detect for early-stage diagnosis, leading to a better prognosis for the patient.
Crossdisciplinary collaboration
This is a project in which state-of-the-art mathematical research for solving large inverse problems meets the clinical practice of medical imaging and brings together faculty from the Department of Mathematics at the SCI school at KTH with faculty from the Department of Biomedical Imaging of the CBH school.
Watch the recorded presentation at the Digitalize in Stockholm 2023 event:
About the project
Objective
This project aims to develop and study a demonstrator of semi-automated online math tutoring by combining a human approach and automated tutoring. The automation will be based on Natural Language Processing analysis of previous tutor-student interactions. To evaluate the semi-automated tutoring concept, we will conduct a randomized controlled trial by randomly assigning students into a treatment group where tutoring is semi-automated and to a control group with only human online tutoring. We will also conduct a thematic comparative analysis of student-automated tutor interaction compared to student-human tutor. The findings will contribute to research on semi-automated tutoring, a scientific area that has received limited attention. The semi-automated approach could greatly impact society because tutors can help more students. The project will serve as an example of beneficially using semi-automation without losing the human touch of tutoring.
Background
It has been known for decades that one-to-one tutoring is a very effective teaching method, although the key challenge is to scale it up. Maths Coach Online (mattecoach.se) has been offering one-to-one tutoring by teacher students to K-12 students using chat and interactive whiteboard since 2009 and has conducted more than 70,000 tutoring conversations. We are transforming Maths Coach Online into a national service, making scaling to a larger volume of students important. Therefore, we want to explore how to support high-quality math learning for as many students as possible by incorporating semi-automated intelligent tutoring.
Crossdisciplinary collaboration
The researchers in the team represent the KTH School of Industrial Engineering and Management, Department of Learning in Engineering Sciences and the KTH School of Electrical Engineering and Computer Science, Department of Intelligent Systems.
Watch the recorded presentation at the Digitalize in Stockholm 2023 event: