3D Computer Vision

Anpei Chen (PostDoc)

We plan to design cutting-edge algorithms to enable fast and generalizable 3D reconstruction, novel view synthesis, and 3D-aware image synthesis by combining advances in computer vision and computer graphics. This project includes three main parts: (1) Geometric modeling and reconstruction are fundamental to compute...

3D Perception, Reconstruction, and Animation of Large-Scale Scenes

Haofei Xu (Ph.D. Student)

This project aims to explore how to enable 3D perception and understanding of large-scale scenes in the following aspects: 1) 3D representation. How to efficiently and effectively represent 3D remains to be an open problem, especially for large-scale scenes. 2) 3D reconstruction and synthesis. Achieving 3D reconstru...

A game theoretic framework for reinforcement learning and imitation learning

Luca Viano (Ph.D. Student)

Reinforcement Learning and Imitation Learning achieved impressive empirical results in recent years. Still, little is known about the theoretical properties of commonly used algorithms. This PhD project aims to use insights from the field of game theory and optimization to develop algorithms with mathematical guaran...

Accountability for AI

Florian Dorner (Ph.D. Student)

With the more and more widespread deployment of Al systems in the real world, they are likely going to have larger and larger impacts on society, including accidents, political destabilization and other negative externalities. To increase developers' accountability for negative societal impacts of their Al systems, ...

Advancing SLAM through Deep Learning

Niclas Vödisch (Ph.D. Student)

A fundamental challenge in the field of mobile robotics is to determine the robot’s pose relative to its environment. Commonly, this is achieved by employing simultaneous localization and mapping (SLAM) systems. While classical methods are well studied and provide reliable solutions to the localization problem, how ...

AI Fairness in Federated Learning

Gergely Dániel Németh (Ph.D. Student)

Federated Learning is a rapidly emerging approach to solve privacy problems regarding Machine Learning applications. Fairness in Artificial Intelligence is also a field that gained awareness in recent years. At the intersection of the two fields, we can find new solutions for problems that exist in centralized machi...

Analyzing Materials through Machine Learning

Vincent Stimper (Ph.D. Student)

Measurement techniques to characterize the properties of materials, such as photoemission spectroscopy, were continuously improved throughout the last decades. This led to a drastic increase of the measured data in terms of size, resolution, and complexity. Analyzing those dataset poses challenges, e.g. processing t...

Application of Neural Networks in Many-Body Physics

Sebastian Sanokowski (Ph.D. Student)

Problems with a large number of interacting particles are ubiquitous in science and key to many future technologies. Recently, the successful application of deep learning methods to such problems led to remarkable progress in statistical mechanics, quantum chemistry and condensed matter physics. We will investigate ...

Attention-guided cross domain visual geo-localization

Gabriele Trivigno (Ph.D. Student)

Photo geolocation is a challenging task since many photos offer only few cues about their location. For instance, an image of a beach could be taken on many coasts across the world. Even when landmarks are present there can still be ambiguity: a photo of the Rialto Bridge could be taken either at its original locat...

Automated Synthesis of Counterfactual Interventions for Explainable and Fair Machine Learning

Giovanni De Toni (Ph.D. Student)

As humans, we make countless decisions every day in many different areas of our lives. Recently, state-of-the-art machine learning models have been used to build automated decision-making systems to enhance human judgment, ranging from accepting or rejecting a job applicant to prescribing medications and treatments....

Bridging Theory and Algorithms for Lifelong Learning

Marco Ciccone (PostDoc)

One of the most ambitious goals of Artificial Intelligence is to provide an agent with the ability to continually learn in an open-ended scenario. Lifelong Learning, or Continual Learning (CL), explicitly focuses on non-stationary or changing environments, where an agent has to complete a series of partially related...

Causal inference under Gaussian Process

Zonghao Chen (Ph.D. Student)

The project aims to incorporate the reliable uncertainty estimate from Gaussian Processes (GP) into causality inference. Recent work on causality inference based on kernel methods have demonstrated both theoretical guarantees and good empirical performances. Therefore, the aim of this project to extend from kernel m...

Causal relationships in nonlinear representation learning

Patrik Reizinger (Ph.D. Student)

The robustness of underlying representations is a key for deploying machine learning systems in real-world applications. This requires both the possibly unique identification of the latent factors themselves and their connections as well. Although the topic is an active research area, there are no satisfactory algor...

Causality in heavy missing data scenarios

Sergio Hernan Garrido Mejia (Ph.D. Student)

Whenever we see the world we rarely observe all the variables of any system of interest at the same time. However, we can create hypotheses of causal relations by combining information of these partially observed scenarios. In this research, we will focus on the question of extracting causal information from scenari...

Combining Probabilistic Inference and Deep Learning

James Urquhart Allingham (Ph.D. Student)

Deep learning is an extremely successful machine learning paradigm that has been the driving force behind the renewed excitement around machine learning over the last seven or so years since AlexNet won the ImageNet LSVRC competition. However, despite the successes of deep learning, it is not without its flaws. One ...

Continual Learning

Mohammad Derakhshani (Ph.D. Student)

Despite the promise of artificially intelligent agents, they are known to suffer from catastrophic forgetting when learning over nonstationary data distributions. Continual learning, also known as life-long learning, was introduced to deal with catastrophic forgetting. In this framework, an agent continually learns ...

Continual Multimodal Leaming

Nico Daheim (Ph.D. Student)

Many NLP tasks require incorporating not only textual data but also other modalities, such as images, speech, or software code. In this project, we aim to build systems that can jointly reason over these different modalities with diverse applications, for example in dialog systems. Furthermore, the information that ...

Continual Reinforcement Learning with associative memories

Thomas Schmied (Ph.D. Student)

The current generation of Deep Reinforcement learning systems is primarily designed to solve one particular task in a single stationary environment. However, the real world is non-stationary and dynamic by nature. For Reinforcement learning agents to be useful under these circumstances, they need to have the ability...

Cost-efficient learning via weak annotators

Hao Qiu (Ph.D. Student)

In this project, we plan to design and analyze new algorithms for sequentially classifying a stream of data points based on a set of costly, noisy, and potentially malicious annotators. Assuming the algorithm obtains noisy labels by adaptively querying selected annotators, we are interested in studying trade-offs be...

Cross-Domain 3D Visual Learning

Antonio Alliegro (Ph.D. Student)

Depth cameras and LiDAR sensors are important tools for agents that need to perceive the world and interact with it, thus 3D data managing and learning algorithms are quickly becoming essential. Deep learning has achieved a significant leap for 2D computer vision, and for 3D applications is currently demonstrating g...

Cross-Lingual Document Summarization

Yifu Qiu (Ph.D. Student)

Automatic document summarization is an area in Natural Language Processing focused on taking input documents and outputting a concise summary of them. Such summaries allow potential readers to sift through a more significant amount of information in a shorter time and focus on the most critical content. Recent work ...

Design and development of computational methods to facilitate the analysis of single-cell multi-omics and single-cell perturbational data

Amir Ali Moinfar (Ph.D. Student)

Single-cell omics allow us to study biological processes at single-cell resolution. The applications of single-cell omic studies range from understanding cellular processes to drug design. We need to design and develop computational models that can handle the complexity and magnitude of data generated by these studi...

Dynamically Evolving Deep Architectures

Rupert Mitchell (Ph.D. Student)

My research focuses on the modification of deep architectures live during training. In contrast to the relatively well explored area of the pure removal of parts of an architecture (i.e. pruning), I am particularly interested in combining this with the addition of parts as well, for example increasing the width or d...

Efficient egocentric vision models for the edge

Gabriele Goletto (Ph.D. Student)

With the advent of machines capable, in principle, to cooperate with humans in performing tasks of daily living, it becomes crucial the development of methods to quickly and easily infer human actions from common data. Egocentric vision is rapidly becoming one of the most promising source of information for this pur...

Explainable AI for Nature Conservation

Diego Marcos (PostDoc)

As Deep Learning gets better at visual tasks, including species identification, the learned reasoning behind its decisions gets increasingly obscure. This is in contrast with the procedures developed by taxonomists, the experts in charge of defining the hierarchy of natural species, for manual species recognition....

Explainable Multimodal Representation Learning

Karsten Roth (Ph.D. Student)

Human perception, but also interaction, is inherently multimodal, i.e. visual and audio information get processed together and the emerging knowledge is communicated through some form of language. In this project, our aim is to learn multimodal representations using artificial neural networks. Our application domain...

Fair Machine Learning Systems

Benedikt Höltgen (Ph.D. Student)

As consequential decisions affecting people's lives are increasingly being delegated to Machine Learning systems, society now needs to grapple with questions of algorithmic fairness. While this is still a fairly new area of research, similar questions have long been discussed both informally and mathematically in ne...

Feedback models for real-time robotic cloth manipulation

Oriol Barbany (Ph.D. Student)

Object rigidity is still one of the most common assumptions in robotic grasping and manipulation. Nevertheless, many daily life objects like cables, plastics, and clothes present non-negligible deformations. Manipulating garments remains a challenging topic due to the high flexibility of textiles and their nearly in...

Fine grained visual analysis using natural language

Vestéinn Snaebjarnarson (Ph.D. Student)

The project is based around analysis of images by use of natural language, in particular for fine grained categorization and analysis. The idea is that the use of natural language will serve both as grounding for direct labelling as well as for comparison to other sources of information. Adaptation of existing gener...

Generative modeling and it's theoretical foundations

Simon Damm (Ph.D. Student)

Generative modeling is a fundamental yet complex task in machine and deep learning. This project aims to investigate and broaden our understanding of the theoretical foundations of (deep) generative models. Our interest is twofold. First, we are interested in theoretical properties of the optimization and training d...

Generative Models and Stochastic Processes

Cameron Stewart (Ph.D. Student)

This project will focus on the further development and analysis of generative models. Examples of such models include Generative Adversarial Networks (GANs), energy-based models, normalising flows, diffusion models, and score-based generative models. Wasserstein gradient flows of various metrics applied to the train...

Geometric Deep Learning

Gabriele Cesa (Ph.D. Student)

Many machine learning tasks come with some intrinsic geometric structure. During my PhD, I will study how to encode the geometry of a problem into neural-network architectures to achieve improved data efficiency and generalization. A particular focus will be given to 3D data and the task of 3D reconstruction, where ...

Grounding symbols in observations via causal, object centric world-models

Anson Lei (Ph.D. Student)

Symbol grounding is a key part of classical AI and harbours significant potential in the ability to perform symbolic and causal reasoning on real-world data. In contrast, machine learning currently operates on raw sensor data and typically without consideration of any symbolic structure in the data. The recent emerg...

Harmonic Analysis for Stochastic Neural Networks

Adeel Pervez (Ph.D. Student)

In this work we explore harmonic analytic techniques to improve the learning capabilities of stochastic neural network models. Stochastic neural networks are widely used in probabilistic modeling for unsupervised and representation learning. Such models are often limited by problems of instability and by the type of...

Holistic 3D Scene Understanding for Self-Driving Cars

Zehao Yu (Ph.D. Student)

Holistic 3D scene understanding plays a critical role in self-driving cars. It involves several sub-tasks, such as geometric layout estimation, object detection, recognition, and tracking. While each sub-task may be solved independently, it would be beneficial to utilize the complementary nature of different sub-tas...

Human Modeling via Reinforcement Learning and Computer Vision

Berat Mert Albaba (Ph.D. Student)

Modeling and imitating human behaviors, actions and decisions from visual data are crucial for the future of AI, yet problematic due to the complexity of humans. In this project, the goal is proposing approaches to model and imitate humans in a generative manner. Thanks to its well adaptation ability to new scenario...

Human-centric Machine Learning in Healthcare

Nina Corvelo Benz (Ph.D. Student)

With machine learning tools being widely used in social and commerce systems and increasingly also to automate processes in other sectors such as medicine, banking, and employment, machine-made decisions are becoming more and more influential to our everyday life. In many of these real-world applications algorithms’...

Improving Deep Learning Explainability with Interpretable Outputs and Representations

Anders Christensen (Ph.D. Student)

Through recent years it has become abundantly clear how powerful machine learning is. Still, despite the empirical success of deep learning models, there is rarely access to the reasoning behind why a model yields a certain output for some input. Achieving more insight into the decision making of such models and dev...

Independent causal mechanisms in machine learning

Julius von Kügelgen (Ph.D. Student)

Due to changes in environment, measurement device, experimental condition, or sample selection bias, the commonly-made assumption of independent and identically distributed (i.i.d.) random variables underlying many machine learning algorithms is often violated in practice. The perspective of causal modelling offers ...

Inductive and Semantic Priors for Categorization in Deep Learning

Tejaswi Kasarla (Ph.D. Student)

An inductive bias of a learning algorithm describes a set of assumptions about the target function independent of training data. Inductive biases play a vital role in the design of machine learning algorithms, consider for example inductive biases for image structures ( e.g., the convolution operator), symmetries (e...

Interactive Multimodal Learning

Haau-Sing Li (Ph.D. Student)

Multimodal learning contains a set of challenging tasks. It requires a deep understanding of all relevant modalities (language, vision, software program) and of the relationships between them. Furthermore, the lack of high-quality datasets increases the difficulty of related tasks. In this project, we will work with...

Interpretable ML for stress forecasting and management

Batuhan Koyuncu (Ph.D. Student)

Modeling temporal stress patterns of people is a challenging task since stress patterns are complex and have a heterogeneous nature. However, it is a crucial task for forecasting and mitigating stress. The objectives of this Ph.D. project are understanding stress causes, forecasting stress levels over time, and prov...

Interpreting NLP models through the lens of cognition and linguistics

Michael Hanna (Ph.D. Student)

Transformer-based models have spurred unprecedented increases in performance on NLP tasks. As these models' performance begins to approach (or even surpass) that of humans, the question of how they achieve this performance becomes crucial. Do these models acquire the same linguistic knowledge, and learn to perform t...

Intrinsic and Invariant Image Decomposition

Xiaoyan Xing (Ph.D. Student)

Intrinsic and invariant information are fundamental components of the image formation process. In fact, the image formation heavily depends on the geometry of the scene, object material, and illumination. To decompose the image into its constituent parts, well-awareness of the geometry information in the scene and t...

Investigating and preventing harmful cultural implications of AI in social media

Piera Riccio (Ph.D. Student)

Social media platforms are reshaping the flow of information in our society, establishing new paradigms of mass communication. Because of their ubiquitous nature, they constantly influence our daily choices, having a widespread impact on different facets of our culture. On such platforms, AI algorithms are massively...

Joint Sentence and Word Alignment

Peiqin Lin (Ph.D. Student)

Word alignment is generally decomposed into two subproblems: first align sentences, then align words on the assumption that only words in aligned sentences can be aligned. This decomposition works well for many genres of parallel corpora, in particular, for parliamentary proceedings and for legal and business text. ...

Learning compact and efficient feature representations

Zhuo Su (Ph.D. Student)

Energy efficient sensing and computing is vital at all levels, from the smallest sensor like the chip to ultra high performance processors and systems like the cloud, especially in the post Moore's Law era. Energy efficient AI enables AI to move beyond the cloud and to reach the edge, which is critical to the progre...

Learning from distributions with kernelized optimal transport

Dimitri Meunier (Ph.D. Student)

The problem of learning functions over spaces of probabilities -- or distribution regression -- is gaining significant interest in the machine learning community. A key challenge behind this problem is to identify a suitable representation capturing all relevant properties of the underlying functional mapping. A pri...

Learning methods for geometry and appearance reconstruction

Božidar Antić (Ph.D. Student)

As XR technologies are gaining more traction, estimating detailed geometry and appearance of scenes and humans is becoming a crucial task in order to have realistic and interactive virtual experiences. The majority of today's methods focus only on reconstructing geometry and simple textures. However, estimation of u...

Learning on Graph Problems in Biology

Paolo Pellizzoni (Ph.D. Student)

Developing machine learning methods for graph-structured data is a theme of paramount importance in bioinformatics. Indeed, graphs are the structure of choice for representing complex objects such as molecules and proteins, which have a central role in many biological problems, as well as for capturing the interacti...

Learning Robot Perception and Localization with Limited Supervision

Julia Hindel (Ph.D. Student)

Comprehensive scene understanding and state estimation are pivotal for achieving reliable robot autonomy in human-centred environments. Yet, the availability of reliable training data undermines the robustness - and thus safety - of such systems, especially in edge cases not or barely covered. This project focuses o...

Learning to solve multiple-view geometry

Diana Sungatullina (Ph.D. Student)

Multi-view geometry is an important field in computer vision and robotics, which provides an understanding to the foundations of the subject. The development of this field has, however, not yet been greatly influenced by the recent advances in machine learning and geometrical machine learning in particular. We aim a...

lmproving Image Synthesis and Manipulation Control with GANs

Edgar Schönfeld (Ph.D. Student)

Among the major challenges of generative adversarial networks (GANs) remain the ability to create images that are indistinguishable from real images and the ability to depict precisely the content specified by the user. To this end, this work proposes methods to improve synthesis quality and maximize control over th...

Machine Learning and Causal Inference to Optimize Genomic Interventions using Disease State Representations

Bowen Fan (Ph.D. Student)

The human genome contains a torrent of information that gives clues not only about human origin, evolution, biological function, but also diseases. The goal of my project aims at developing novel machine learning techniques to better understand the complex genomic data and also other forms of data that can represent...

Machine Learning for Biological Network Analysis

Giulia Muzio (Ph.D. Student)

The main objective of my project is to develop methods for network-based genome-wide association studies (GWAS) that combine computational efficiency, statistical power and interpretability, thereby enabling the discovery of biological pathways underlying complex phenotypic traits. GWAS aim to identify statistical ...

Machine Learning for Single-Cell Omics

Merel Kuijs (Ph.D. Student)

In spite of the massive effort in ensuring drug safety and efficacy, some drugs fail to achieve a therapeutic effect and others have severe side effects. A personalized approach to medicine is needed to maximize every patient’s chance of recovery and relieve pressure on healthcare systems, but data to inform precisi...

Machine Learning for the Fusion of Remote Sensing and Tweets Data for Green Space Analysis

Mohamed Ibrahim (Ph.D. Student)

There is increasing evidence that people with higher access to urban green spaces have better mental health and well-being. This project aims to examine the impact of urban green spaces on mental health, and what features play the biggest role. The combination of airborne or space-borne remote sensing images and geo...

Mathematical aspects of variational quantum machine learning models

Elies Gil-Fuster (Ph.D. Student)

Recent years have seen the birth and growth of quantum machine learning, a new discipline which uses quantum computers for data analysis. Within this field, many heuristics have been proposed and to some extent individually tested. Yet, there is no truly holistic framework to study and characterize the essence and q...

Methods for Robust Feature Learning

Sindy Löwe (Ph.D. Student)

The representation of objects and their relations lies at the center of our ability to reason and to adapt to new environments. In my PhD, I want to develop algorithms that learn object-centric representations from low-level perceptual inputs in an unsupervised fashion.

Modeling Humans in the Scene with Compositional and Controllable Neural Representations

Zijian Dong (Ph.D. Student)

This project aims to develop generative 3D models for controllable 3D synthetic humans in a scene. The first part of the project is to generate one animatable neural human avatars given only a few RGB/RGBD images of a person from different views. The second part of the project is to represent a scene with compositio...

Modeling the molecular response of single cells to drug perturbations

Artur Szalata (Ph.D. Student)

Biological single-cell sequencing data can be used to learn about cellular phenotypes, developmental trajectories, and disease. It is increasingly utilized to measure a cell's molecular state across perturbations. To better understand a cellular system, we develop a machine learning model that can accurately predict...

Modular Language and Domain Adaptation of Language Models

Indraneil Paul (Ph.D. Student)

This project aims to mitigate model development and maintenance costs at all points along its lifecycle by improving modularity. We seek to improve efficiency using modular additions to the network and do so effectively in a manner that does not incur the cost of catastrophic interference. Such methods can, in time,...

Multi-Task Learning for Medical Image Analysis and Data Mining

Jiayi Shen (Ph.D. Student)

Multi-task learning is a fundamental learning paradigm for machine learning, which aims to simultaneously solve multiple related tasks to improve the performance of individual tasks by sharing knowledge. The crux of multi-task learning is to explore task relatedness to improve each individual task, which is non-triv...

Multimodal video learning

Yunhua Zhang (Ph.D. Student)

Video streams consist of multiple modalities, e.g., RGB frames, optical flow, audio. Their natural correspondence provides rich semantic information to achieve effective multi-modal perception and learning. In this project, our aim is to understand the video content by analyzing multiple modalities and decide which ...

Neuro-Symbolic Reinforcement Learning for Programming Education

Manh Hung Nguyen (Ph.D. Student)

This research project is in the area of Reinforcement Learning and AI for Programming Education. Given that programming tasks are open-ended and conceptual, novice students often struggle when learning to program. AI-driven tutors hold great promise in automatically assisting and teaching novice students. The projec...

On Learning Techniques for Optimal Control of Legged Robots

Daniel Felipe Ordoñez Apraez (Ph.D. Student)

In this project, we will focus on the integration of data-driven methods to model-based optimal-control pipelines that formulate the optimization process as a trajectory optimization. The objective is to find application areas for data-driven methods to improve the control pipeline of a legged robot. The tentative a...

Out-of-distribution Generalization

Zehao Xiao (Ph.D. Student)

Out-of-distribution (OoD) generalization is a common task in machine learning. Deep neural networks have exhibited remarkable performance over various computer vision and machine learning tasks. However, they struggle and lack robustness when the test data are in different distributions from the training data. These...

Persistent Impacts in Vegetation and Extreme Events

Tristan Williams (Ph.D. Student)

Persistence is an important characteristic of many complex systems in nature and of the Earth system in particular. The concept is rather elusive but related to how long the system remains at a certain state before changing to a different one. Characterising persistence in the terrestrial biosphere is very relevant ...

Preventing the spread of misinformation

Peter Ebert Christensen (Ph.D. Student)

As a part of doing research communication between different partners is key, both when working with teams and writing papers. Research into what people are actually saying, e.g. their narratives surrounding a topic, despite the unknown degree of the truth of it. Traditionally the focus has been on fact checking with...

Privacy-preserving data sharing via probabilistic models

Joonas Jälkö (Ph.D. Student)

Widespread sharing of data would facilitate rapid progress in data science. However, due to privacy constraints, sensitive data cannot be made public. My research aims to learn a generative model from the sensitive data under strict privacy guarantees from differential privacy. The generative model is then used to d...

Probabilistic Reasoning with Neural Networks Beyond Weight Space

Javier Antorán (Ph.D. Student)

Neural networks are a flexible class of models that have soared in popularity due to their scalability to large amounts of data and computation. The fact that these models are able to represent a very broad range of functions makes performing Bayesian inference in them challenging. Most existing approaches focus on ...

Quantifying uncertainty in deep learning models

Lukas Aichberger (Ph.D. Student)

Predictions made by machine learning models urge to be reliable. Understanding if a prediction is under-confident or falsely over-confident is critical for deploying machine learning, especially in real-world applications. However, extracting models from data is inseparably connected with uncertainty, due to the int...

Reinforcement Learning for Education

Skander Moalla (Ph.D. Student)

Sequential student-teacher interactions are a central component of education (ED). Effective interactions significantly accelerate learning, however, they cannot be massively scaled when requiring a human teacher in the loop, and otherwise, have been hard to provide when the teaching is exclusively done by a compute...

Representation Learning over Multi-Relational Graphs

Yihong Chen (Ph.D. Student)

One of the main characteristics of human intelligence is the ability to reason over relationships among various entities. Multi-relational graphs, specifically knowledge graphs, are an excellent way to capture such relational knowledge. Good representation learning over multi-relational graphs fosters downstream app...

Representation Learning with Deep Generative Models

Andrea Dittadi (Ph.D. Student)

Learning useful representations from data with little or no supervision is a key challenge in artificial intelligence. Firstly, while labeled data is typically expensive, vast amounts of unlabeled data are available. Secondly, although the usefulness of a representation depends on the downstream task, it should be p...

Sample Efficient Reinforcement learning via Language Abstractions

Fabian Paischer (Ph.D. Student)

Deep Reinforcement learning has gained plenty of attention recently through mastering highly complex games, such as StarCraft II. Many difficult problems arise in such games, for example dealing with partial observability, or continuous state and action spaces. To cope with such problems, current algorithms require ...

Semantic Labeling of Multisensory 3D Point Clouds

Yuxing Xie (Ph.D. Student)

Benefiting from the unprecedented technology development of sensors, platforms and algorithms for 3D data acquisition and generation, point clouds are becoming more significant and accessible than before. In addition to widely-used LiDAR point clouds, satellite stereo imagery- and InSAR- based 3D data also cannot be...

Statistical Inference in Deep Science-Aware Models

Kai-Hendrik Cohrs (Ph.D. Student)

Science-aware and science-guided machine learning attempts to combine knowledge of first principles and machine learning. One way of doing so is using explicit equations from theoretical knowledge and estimate parameters or latent states with machine learning approaches. This promises scientifically interpretable re...

Studying generalisation using prompting for data augmentation

Guillem Ramírez Santos (Ph.D. Student)

For the last five years, a new level of performance in NLP has been achieved by leveraging unsupervised training of large pretrained language models (PLM) and then fine-tuning on a downstream task. However, this is changing recently, as the community is shifting its attention towards prompting. The success of prompt...

Temporal Causality in Machine Learning

Phillip Lippe (Ph.D. Student)

The field of Causality is concerned with finding and understanding the effect that a change of a variable has on other variables in an environment. Meanwhile, current machine learning approaches typically rely on correlations in data but lack the clear separation of cause and effect. This can lead to problems when t...

Tiny Adaptive Deep Learning

Francesco Tonini (Ph.D. Student)

Deep learning has significantly improved the state of the art for several tasks in computer vision and robotic perception. However, a major obstacle in adopting deep models in real world applications is the fact that their performances significantly drop when there is a shift in data distributions across training an...

Towards Bayesian Neural Model Selection

Alexander Immer (Ph.D. Student)

Model selection and comparison, the task of choosing the optimal model for a given data set, is a central problem in machine learning. Model selection is particularly challenging for deep neural networks because of their ability to represent arbitrary functions and their enormous size and overparameterization. For t...

Towards joint reasoning of humans and scenes for character animation and performance capture

Guoxing Sun (Ph.D. Student)

Mr. Sun will explore novel and neural algorithms and representations, which in conjunction with new datasets will allow to synthesize novel motions and interactions with objects in the scene. In contrast to classical computer animation, one goal of his thesis is to simplify the user input to enable the usage by non-...

Towards Semantic Speech Processing

Lasse Borgholt (Ph.D. Student)

Human spoken language understanding relies heavily on contextual information. If we consider a single word, the context might simply be the words surrounding it. These words provide important clues for the listener to accurately recognize and understand the word of interest. This word may be hard to recognize becaus...

Trustworthy Self-Attentive Models for Visual-Semantic Understanding