Search Results for author: Mario Michael Krell

Found 12 papers, 4 papers with code

Tuple Packing: Efficient Batching of Small Graphs in Graph Neural Networks

no code implementations14 Sep 2022 Mario Michael Krell, Manuel Lopez, Sreenidhi Anand, Hatem Helal, Andrew William Fitzgibbon

However, the sizes of small graphs can vary substantially with respect to the number of nodes and edges, and hence the size of the combined graph can still vary considerably, especially for small batch sizes.

Classifier Transfer with Data Selection Strategies for Online Support Vector Machine Classification with Class Imbalance

no code implementations10 Aug 2022 Mario Michael Krell, Nils Wilshusen, Anett Seeland, Su Kyoung Kim

Adding all and removing the oldest samples results in the best performance, whereas for smaller drifts, it can be sufficient to only add potential new support vectors of the SVM which reduces processing resources.

EEG online learning

Efficient Packing: Towards 2x NLP Speed-Up without Loss of Accuracy for BERT

no code implementations29 Sep 2021 Matej Kosec, Sheng Fu, Mario Michael Krell

The shortest-pack-first histogram-packing (SPFHP) algorithm determines the packing order for the Wikipedia dataset of over 16M sequences in 0. 02 seconds.

NanoBatch Privacy: Enabling fast Differentially Private learning on the IPU

no code implementations24 Sep 2021 Edward H. Lee, Mario Michael Krell, Alexander Tsyplikhin, Victoria Rege, Errol Colak, Kristen W. Yeom

We also provide two extensions: 1) DPSGD for pipelined models and 2) per-layer clipping that is 15x faster than the Opacus implementation on 8x A100s.

Efficient Sequence Packing without Cross-contamination: Accelerating Large Language Models without Impacting Performance

1 code implementation NeurIPS 2021 Mario Michael Krell, Matej Kosec, Sergio P. Perez, Andrew Fitzgibbon

We show in this paper that the variation in sequence lengths in common NLP datasets is such that up to 50% of all tokens can be padding.

Hardware-accelerated Simulation-based Inference of Stochastic Epidemiology Models for COVID-19

1 code implementation23 Dec 2020 Sourabh Kulkarni, Mario Michael Krell, Seth Nabarro, Csaba Andras Moritz

The statistical inference framework is implemented and compared on Intel Xeon CPU, NVIDIA Tesla V100 GPU and the Graphcore Mk1 IPU, and the results are discussed in the context of their computational architectures.


Accelerating Simulation-based Inference with Emerging AI Hardware

1 code implementation12 Dec 2020 Sourabh Kulkarni, Alexander Tsyplikhin, Mario Michael Krell, Csaba Andras Moritz

As a proof-of-concept, we demonstrate inference over a probabilistic epidemiology model used to predict the spread of COVID-19.


A First Step Towards Distribution Invariant Regression Metrics

no code implementations10 Sep 2020 Mario Michael Krell, Bilal Wehbe

We show on synthetic and robotic data in reproducible experiments that classical metrics behave wrongly, whereas our new metrics are less sensitive to changing distributions, especially when correcting by the marginal distribution in $X$.

Learning of Multi-Context Models for Autonomous Underwater Vehicles

no code implementations17 Sep 2018 Bilal Wehbe, Octavio Arriaga, Mario Michael Krell, Frank Kirchner

Multi-context model learning is crucial for marine robotics where several factors can cause disturbances to the system's dynamics.

General Classification

Data Augmentation for Brain-Computer Interfaces: Analysis on Event-Related Potentials Data

no code implementations9 Jan 2018 Mario Michael Krell, Anett Seeland, Su Kyoung Kim

On image data, data augmentation is becoming less relevant due to the large amount of available training data and regularization techniques.

Data Augmentation

Cannot find the paper you are looking for? You can Submit a new open access paper.