Jekyll2023-09-07T10:03:40+00:00http://goschjann.github.io/feed.xmlintro jann goschenhofer2019-04-24T22:00:00+00:002019-04-24T22:00:00+00:00http://goschjann.github.io/pwpd_paper<p>I pre-published the results of my master’s thesis on <em>Wearable-based Parkinson’s Disease Severity Monitoring using Deep Learning</em>. Read the paper <a href="https://arxiv.org/abs/1904.10829">here</a>.</p>goschjann

I pre-published the results of my master’s thesis on Wearable-based Parkinson’s Disease Severity Monitoring using Deep Learning. Read the paper here.

2019-03-17T22:00:00+00:002019-03-17T22:00:00+00:00http://goschjann.github.io/dadada2019<p>I had the chance to talk about my Master’s thesis at the Datageeks Data Day (DADADA) 2019 in Munich. Awesome event, awesome community and awesome discussions on quantification of uncertainty in machine learning.</p> <ul> <li>read <a href="https://goschjann.github.io/ddd_talk/">the slides</a> (use fullscreen mode for the optimal experience)</li> <li>watch <a href="https://www.youtube.com/watch?v=pP6T28otgfk">the talk</a></li> </ul>goschjann

I had the chance to talk about my Master’s thesis at the Datageeks Data Day (DADADA) 2019 in Munich. Awesome event, awesome community and awesome discussions on quantification of uncertainty in machine learning.

2018-05-16T10:00:00+00:002018-05-16T10:00:00+00:00http://goschjann.github.io/char_cnn<p>Small project on text analysis via CNN’s within a uni seminar. Using an alphabet-based character-level dummy-encoding allows for the application of 1D Cnns for classification of yelp reviews. Based on work from <a href="https://arxiv.org/abs/1509.01626">Zhang, Zhao and LeCunn (2015)</a>. Also included Local Interpretable Model-Agnostic Explanations (<strong>LIME</strong>) by <a href="https://homes.cs.washington.edu/~marcotcr/blog/lime/">Marco Ribeiro</a> to make sense out of complex decision makers such as CNNs.</p> <ul> <li>check <a href="https://github.com/Goschjann/charCnn">the repo</a></li> <li>read <a href="https://goschjann.github.io/charCnn/">the slides</a></li> </ul>goschjann

Small project on text analysis via CNN’s within a uni seminar. Using an alphabet-based character-level dummy-encoding allows for the application of 1D Cnns for classification of yelp reviews. Based on work from Zhang, Zhao and LeCunn (2015). Also included Local Interpretable Model-Agnostic Explanations (LIME) by Marco Ribeiro to make sense out of complex decision makers such as CNNs.

2018-01-01T09:30:00+00:002018-01-01T09:30:00+00:00http://goschjann.github.io/road_segmentation<p>Together with <a href="https://de.linkedin.com/in/niklas-klein">Niklas Klein</a>, I was working on a road segmentation (in: satellite image, out: road map) using the infamous U-Net as part of our master’s curriculum.</p> <ul> <li>read <a href="https://github.com/Goschjann/road_segmentation_project/blob/master/slides/talk_datageeks_meetup.pdf">the slides</a> from a talk about it at a Meetup in 2018</li> <li>check <a href="https://github.com/Goschjann/road_segmentation_project">the repo</a> and reproduce our results with public data</li> </ul>goschjann

Together with Niklas Klein, I was working on a road segmentation (in: satellite image, out: road map) using the infamous U-Net as part of our master’s curriculum.