Hi. I'm Jonas Klaff.

I am a Machine Learning Research Scientist and recently obtained my M.Sc. in Computer Science with honors ("ausgezeichnet") at the University of Bremen with an official focus on Artificial Intelligence, Cognition and Robotics. In my thesis (pdf, slides) I focused on using Active Learning and Transfer Learning for Human Motion Analysis and 3D Perception and in the challenging field of Combat Sports.

I am currently working on Domain Adaption and Knowledge Distillation in the field of Natural Language Processing & Understanding and Automatic Speech Recognition.

My prior work is mainly focussed on Deep Learning for Natural Language Processing tasks, namely Sequence Classification & Tagging, Argumentation Mining, Topic Modelling and Event&Entity Extraction. But I also took little detours (to be published at IEEE SMC 2021 ) to the fascinating field of Machine Learning on Biosignals, e.g. in Brain-Computer-Interfaces where we a classified human attention using EMG data.

Please check out my:
Specialization Recent projects Publications

My Specialization.

My main focus is on Machine Learning and Artificial Intelligence engineering and development, but I am also capable of presenting my results as a Web- or Android App.

Natural Language Processing

Automatic Speech Recognition

Computer Vision

Publications.

  1. G. Skitalinskaya, J. Klaff, and M. Spliethoever. CLEF ProtestNews lab 2019: Contextualized Word Embeddings for Event Sentence Detection and Event Extraction. In CLEF (Working Notes) 2019, Lugano, Switzerland, September 9-12, 2019
  2. M. Spliethöver, J. Klaff, and H. Heuer. Is It Worth the Attention? A Comparative Evaluation of Attention Layers for Argument Unit Segmentation. In Proceedings of the 6th Workshop on Argument Mining, pages 74–82, Florence, Italy, Aug.2019. Association for Computational Linguistics. Code
  3. G. Skitalinskaya, J. Klaff, and H. Wachsmuth. Learning from Revisions: Quality Assessment of Claims in Argumentation at Scale. In The 16th Conference of the European Chapter of the Association for Computational Linguistics, Kyiv, Ukraine, April 2021. Code
  4. L-M. Vortmann, J. Klaff, T. Urban and F. Putze. SSVEP-Aided Recognition of Internally and Externally Directed Attention from Brain Activity. In 2021 IEEE International Conference on Systems, Man, and Cybernetics (SMC), Melbourne, Australia, October 2021.

Here's an overview of my recent projects.

This is a non exhaustive list of projects I did on my own or was involved in.

3DPunch (Master's Thesis)

My master's thesis "3DPunch: Deep Learning for 3D Pose Estimation from Monocular Video for Human Motion Analysis in Olympic Boxing" based on Computer Vision and Human Pose Estimation for Activity Classification in Boxing Fights. For a deep dive please check out the pdf and slides

Python
PyTorch
TensorFlow
Docker

Oktober 2020 - June 2021

Argument Quality Assessment at Scale

In this paper, we study claim quality assessment irrespective of discussed aspects by comparing different revisions of the same claim. We compile a large-scale corpus with over 377k claim revision pairs of various types from kialo.com, covering diverse topics. We then propose two tasks: (a) assessing which claim of a revision pair is better, and (b) ranking all versions of a claim by quality. Checkout the Paper as well

Python
TensorFlow
ScikitLearn
Docker

May 2020 - Oktober 2020

Brain-Computer-Interfaces

Planing and carrying out an EEG-based study as well as training an LDA-Classifier to classify internal or external attention based on EEG-Signals. Strong focus on visualisation and analyzation of Bio-Signals as well as on statistical machine learning.

Python
MNE-Python
ScikitLearn

September 2020 - Oktober 2020

Geometrical Reconstruction (3D)

Data-driven Camera-Calibration and visualization of a free kick circle of a fixed radius in monocular 2D images by projecting from 2D to 3D and applying the Least-Squares Algorithm.

GitHub   nbviewer

Python
TensorFlow

June 2020 - August 2020

Strawberry Detection in the Wild

Implementing a custom head for Mobile Net to detect (ripe) strawberries in a custom dataset, with the general requirement to keep a minimal network architecture.

GitHub   nbviewer

Python
TensorFlow

May 2020 - June 2020

Bremen Big Data Challenge 2020

Winning team of the Bremen Big Data Challenge with a Deep Convolutional Neural Network for detecting different Activities based on Motion Capture Data.

Detailed Task Description   Leaderboard

Python
TensorFlow
ScikitLearn

April 2020 - May 2020

Is It Worth the Attention?

A Comparative Evaluation of Attention Layers for Argument Unit Segmentation. First author (equal contribution) of the Paper at the ACL2019. Evaluation Attention Layers in combination with Bidirectional LSTM RNNs for a NLP Task.

Paper   Code

Python
TensorFlow (Keras)
ScikitLearn

January 2019 - June 2019

Argumentative Sentence Detection

Bachelor thesis on analyzing the effect of increased complexity in terms of model size and input representation size, on the performance of the given machine learning model. I evaluated a Decision Tree, a Support Vector Machine and a Bidirectional LSTM RNN for different input representations such as Bag-of-Words and Word Embeddings.

Python
TensorFlow
ScikitLearn

July 2019 - November 2019

ForCarster

Bachelor Project on predicting the availability of FreeFloating CarSharing Vehicles with a Random Forest Classifier. Project was in collaboration with other students. My role was the data-preparation, -transformation and training/evaluating different machine learing algorithms.

Python
ScikitLearn

Oktober 2017 - July 2018

Work projects.

This is an overview of my employements in computer science related jobs.

Global Welfare State Information System

Student Asssistent - Software Development & DevOps

I am currently employed in an interdisciplinary research project to co-create an interactive web-based information system with social scientists that provides visual analysis tools to support social policy research.

My role is to develop the information system within a group of 3 computer science researchers and another student assistant, as well as I set up the coninuous integration workflow and maintain the actively used system in production.

Ruby (on Rails)
React
Javascript
Monit
PostgreSQL
Ubuntu
Capistrano
Webpacker

July 2018 - now

Trauerland - Zentrum für trauernde Kinder und Jugendliche e.V.

Student Employee - IT-Management

During my Bachelor's degree I was supporting the amazing work of the non-profit organization Trauerland, as a Student Employee.

My role was to assist the launch of a Customer Relationship Management System, implementing new submodules to adapt the system to the organizations workflow, organizing and leading workshops to train employees and identify data-sources, implement reporting tools for data analysis as well as maintaining and modernizing the general IT-infrastructure across three different locations.

CASgenesisWorld (CRM)
Windows Server
SQL
Tobit David3 Mail Server
Pascom VoIP Software

December 2015 - October 2018

Find me on ...