The following projects have received funding from Chalmers AI Research Centre.
Projects started 2020
Deep Learning and likelihood-free Bayesian inference for intractable stochastic models
We construct new deep neuronal networks (DNNs) to learn the
parameters of complex stochastic dynamical models that do not have
tractable likelihood functions. Specifically, we leverage the expressive
approximation power of our DNNs to extract essential information from
time-series data, both Markovian and not-Markovian, and then learn model
parameters using likelihood-free methodology, such as approximate
Bayesian computation. Special (though not exclusive) focus is directed
to stochastic differential equation models and state space models
(SSMs), where SSMs represent noisy observations of a latent Markovian
process. The result will be a flexible plug-and-play machine learning methodology, allowing inference for complex stochastic models.
Applicant: Umberto Picchini, Department of Mathematical Sciences, Chalmers
Energy-based models for supervised deep neural networks and their applications
Despite deep learning-based methods being the state-of-the-art in
many AI-related applications, there is a lack of consensus of how to
understand and interpret deep neural networks in order to reason about
their strengths and weaknesses. Energy-based models in machine learning
have a long tradition as a framework to learn from unlabeled data, i.e.
unsupervised learning. Recently, it has been shown that supervised
learning of deep neural networks using the back propagation method is a
limiting case of a suitably defined approach for learning energy-based
models using a so-called contrastive loss. This connection is the basis
for our interest in a tighter connection between deep learning and
energy-based models.
Applicants: Christopher Zach, Department of Electrical Engineering, Chalmers, and Morteza Haghir Chehreghani, Department of Computer Science and Engineering, Chalmers
Mechanisms for secure and private machine learning
We envision secure and privacy-preserving machine learning algorithms
for artificial intelligence applications in everyday life, that can
provide confidentiality and integrity guarantees. In particular, we aim
to:
- Safeguard the privacy of individuals that participate by either
(a) providing their data to build the system, or (b) being end-users of
the system.
- Safeguard the integrity of the system by (a) ensuring its
robustness to adversarial inputs (b) cryptographically limiting the
possible points of adversarial manipulation.
Applicants: Aikaterini Mitrokotsa, Department of Computer Science and Engineering, Chalmers, and
Christos Dimitrakakis, Department of Computer Science and Engineering, Chalmers
Stochastic continuous-depth neural networks
We will advance the understanding of deep neural networks through the
investigation of stochastic continuous-depth neural networks. These can
be thought of as deep neural networks (DNN) composed of infinitely many
stochastic layers, where each single layer only brings about a gradual
change to the output of the preceding layers. We will analyse such
stochastic continuous-depth neural networks using tools from stochastic
calculus and Bayesian statistics. From that, we will derive practically
relevant and novel training algorithms for stochastic DNNs with the aim
to capture the uncertainty associated with the predictions of the
network.
Applicant: Moritz Schauer, Department of Mathematical Sciences
VisLocLearn - Understanding and Overcoming the Limitations of Convolutional Neural Networks for Visual Localization
Visual localization is the problem of estimating the position and
orientation from which an image was taken with respect to the scene. In
other words, visual localization allows an AI system to determine its
position in the world through a camera. Understanding why current
approaches fail and proposing novel approaches that are able to
accurately localize a camera are problems of high practical relevance.
This is the purpose for the proposed project, VisLocLearn.
Applicant: Torsten Sattler, Department of Electrical Engineering
Projects started 2019
3D Perception and Prediction Based on Deep Learning
The overall purpose of the project is to basic framework for combining perception, prediction, and control to optimize system level performance. Deep learning algorithms in the vision system will be developed in connection to dynamic models of the pedestrians, where the uncertainty of the predictions will be crucial for the control performance.
PI: Fredrik Kahl, Electrical Engineering, Chalmers
AI-Integromics: Interpretable deep modeling for Systems Biology
We aim to provide AI solutions for Integromics that are effective and interpretable. To this end, an AI-Integromics approach should predict phenotypes for in-house data, attribute predictive strength to the multi-omics source data, yet also provide a mapping to extensive public data resources to facilitate drug re-purposing.
PI: Rebecka Jörnsten, Mathematical Sciences, Chalmers
Deep Reinforcement Learning: Principles and Applications in Cognitive Science and Networks
The overall project goals are to carry out basic research in Deep Reinforcemnet Learning (DRL) which is one of the most powerful and exible frameworks available today for general purpose learning, and to apply it in two areas: (a) fundamental problems in cognitive science in collaboration with the world's leading group in this area at UC Berkeley and (b) network policies for applications at Ericsson.
PI: Devdatt Dubhashi, Computer Science and Engineering, Chalmers
Over the last decade, deep-learning algorithms have dramatically improved the state of the art in many machine-learning problems, including computer vision, speech recognition, natural language processing, and audio recognition. Despite their success, however, there is no satisfactory mathematical theory that explains the functioning of such algorithms. Indeed, a common critique is that deep-learning algorithms are often used as black box, which is unsatisfactory in all applications for which performance guarantees are critical (e.g., traffic-safety applications). The purpose of this project is to increase our theoretical understanding of deep neural networks. This will be done by relying on tools of information theory and focusing on specific tasks that are relevant to computer vision.
PI: Giuseppe Durisi, Electrical Engineering, Chalmers
The development and evolution of software is quite fundamentally changing. Whereas earlier software was exclusively built based on requirements, in more recent years we see that there are three distinct approaches to the creation of software, i.e. requirements-driven, outcome-driven and AI-driven.
PI: Jan Bosch, Computer Science and Engineering, Chalmers
Verification of Machine Learning Algorithms (Vermillion)
Investigate if new methods of testing of cyber-physical systems being developed in our VR Researchv Environment [SyTec,CSE18] can be adapted to work on neural networks. Study applications in ML based Natural language translation (where errors are everywhere, and grouping them is what is needed). Study testing and formal verification of ML based autonomous driving in collaboration with RiSE Viktoria. Study testing in the context of machine learning based software development (see Erik Meijer’s recent talk
at Chalmers; Meijer has agreed to be an advisor on the project). The postdoc will likely concentrate on one of these application domains, depending on their background.The long term goal is to develop methods to design neural networks and machine learning based software with guaranteed properties.