View All Pages

Profile of TAs

To help you decide with whom you should book a help session, here is a list of the specific expertise our TAs have:

Sofia Broomé

Brief Bio: Phd student at RPL working on detecting pain expressions in horses from video data.

Deep learning software packages:

  • TensorFlow
  • Keras

Deep learning methodologies, architectures and application fields:

  • RNNs/LSTMs for sequential image data. 
  • But basically anything that can model time dependencies in vision: convolutional LSTMs, 3D convolutions, CNNs with RNNs stacked on them, etc. 

 

Taras-Svitozar Kucherenko

Brief Bio: PhD at RPL supervised by Hedvig Kjellström. 

He was a TA for DD2424 in 2017 and 2018. (and the students thought he was great added by Josephine) 

Deep learning software packages:

  • TensorFlow
  • Keras

Deep learning methodologies, architectures and application fields:

  • Taras has implemented NN and backprop from scratch during his bachelor studies. He applied it to problems in cryptography.
  • During the first year of his PhD Taras has applied Deep Learning to the problem of Missing Markers Reconstruction:  https://arxiv.org/abs/1803.02665 Links to an external site.

Yue Liu

Brief Bio:  Phd student working on deep learning in medical image analysis.
Deep learning software packages:
  • TensorFlow 
Deep learning methodologies, architectures and application fields :
  • Training a deep learning network to access breast cancer risk.
  • Efficient image classification while preserving good accuracy, with recurrent residual networks improved by feature transfer. 

 

Matteo Gamba

Brief Bio: PhD student at RPL, working on characterizing memorization in deep networks.
He was TA for the course DD2437 Artificial Neural Networks and Deep Architectures in 2017, 2018 and 2019.
Deep learning software packages:
  • PyTorch
  • Caffe (C++)
Deep learning methodologies, architectures and application fields :
  • CNNs for computer vision -- the main focus of my research is building an experimental framework for studying memorization and generalization in deep convolutional networks
  • focus on theoretical aspects of deep learning rather than on specific datasets
  • as part of his undergraduate studies, Matteo implemented a privacy-preserving variant of SGD based on homomorphic cryptography with C++ and Caffe.

 

Marcus Klasson

Brief Bio: PhD student in computer vision at RPL.
Deep learning software packages:
  • TensorFlow 
  • Keras
Deep learning methodologies, architectures and application fields :
  • Currently I am interested in problems that lies at the intersection of computer vision and natural language processing, e.g. image caption generation and visual question answering (VQA)
  • Very excited about generative models, including variational autoencoders and generative adversarial networks, and whenever I am brave enough I like to read about Bayesian deep learning

 

Óttar Guðmundsson

Brief Bio: M.Sc. student Software Engineering of Distributed Systems 
Deep learning software packages:
  • TensorFlow 
  • Keras
Deep learning methodologies, architectures and application fields :
  • Been working with Neural Networks for the past two years. Last year in DD2424 we combined CNN with an Extreme Learning machine for spoken verb recognition. We also looked at the production part, where we connected the software as a service to a slideshow, having the final presentation partly controlled to what we were saying instead of skipping the slides manually.
  • I'm currently working on my master thesis which is about using Gating networks for decision fusion between weak experts at the Karolinska Institutet. It trains on multimodal data.
  • Really like generative models such as generative adversarial networks, but I've been creating album covers using a GAN trained on pop albums.

 

Yonk Shi

Brief Bio: M.Sc Student in Machine Learning  
Deep learning software packages:
  • TensorFlow 
  • Keras
Deep learning methodologies, architectures and application fields :
  • Currently I am writing my thesis on topics related to deep reinforcement learning and disentangled VAEs. 
  • Previously I have worked with GANs and residual networks. Text to image synthesis with GAN. 

 

Thomas Gaddy

Brief Bio: M.Sc Student in Machine Learning  
Deep learning software packages:
  • PyTorch 
  • TensorFlow
Deep learning methodologies, architectures and application fields :
  • I am currently writing my thesis on topics related to unsupervised learning of interpretable representations at Tobii AB.

 

Zehang Weng

Brief Bio: M.Sc Student in Machine Learning  
Deep learning software packages:
  • PyTorch 
  • TensorFlow
Deep learning methodologies, architectures and application fields :
  • I am currently doing my thesis project on topics related to dataset curation using deep learning at Scilifelab.
  • Previously I have worked with object detection using R-CNN framework.

 

Borja Rodríguez Gálvez

Brief Bio: M.Sc Student in Machine Learning  
Deep learning software packages:
  • PyTorch
  • Keras (with tensorflow backend)
Deep learning methodologies, architectures and application fields :
  • Currently writing my thesis on theoretical bounds for machine learning and information theory. Particular emphasis is paid to the information bottleneck and deep learning.
  • Have experience with VAEs, GANs, GMMNs.

Federico Baldassarre Links to an external site.

Brief Bio: PhD student in Deep Learning at RPL
Deep learning software packages:
  • PyTorch
  • Keras and TensorFlow in the past
Deep learning methodologies, architectures and application fields :
  • Very excited about everything concerning computer vision, from dense detections, action recognition, pose estimation, image generation etc.
  • Currently working on methods for interpreting the predictions of deep neural networks