Academic interests
- Machine learning
- Evolutionary optimization
- Automatic programming
Teaching
- ITL25019 Big Data: Storage and Processing - course leader
- Supervisor for master projects within machine learning
- Supervisor for projects within the subject ITI49114 Research project
Background
See my website: www.geitle.no
Also chairman of Future Then AS
Tags:
The Digital Society
Publications
-
Nareklishvili, Maria & Geitle, Marius
(2024).
Deep Ensemble Transformers for Dimensionality Reduction.
IEEE Transactions on Neural Networks and Learning Systems.
ISSN 2162-237X.
36(2),
p. 2091–2102.
doi:
10.1109/TNNLS.2024.3357621.
Full text in Research Archive
Show summary
We propose deep ensemble transformers (DETs), a fast, scalable approach for dimensionality reduction problems. This method leverages the power of deep neural networks and employs cascade ensemble techniques as its fundamental feature extraction tool. To handle high-dimensional data, our approach employs a flexible number of intermediate layers sequentially. These layers progressively transform the input data into decision tree predictions. To further enhance prediction performance, the output from the final intermediate layer is fed through a feed-forward neural network architecture for final prediction. We derive an upper bound of the disparity between the generalization error and the empirical error and demonstrate that it converges to zero. This highlights the generalizability of our method to parameter estimation and feature selection problems. In our experimental evaluations, DETs outperform existing models in terms of prediction accuracy, representation learning ability, and computational time. Specifically, the method achieves over 95% accuracy in gene expression data and can be trained on average 50% faster than traditional artificial neural networks (ANNs).
-
Shahraki, Amin; Geitle, Marius & Haugen, Øystein
(2020).
A comparative node evaluation model for highly heterogeneous massive‐scale Internet of Things‐Mist networks.
Transactions on Emerging Telecommunications Technologies.
ISSN 1124-318X.
31(12),
p. 1–28.
doi:
10.1002/ett.3924.
Full text in Research Archive
Show summary
Internet of Things (IoT) is a new technology that is driving the connection of billions of devices around the world. Because these devices are often resource‐constrained and very heterogeneous, this presents unique challenges. To address some of these challenges, new paradigms of Edge and Fog are emerging to bring computational resources of the IoT networks from remote devices like cloud closer to the end‐devices. Mist computing is a new paradigm that attempts to make use of the more resource‐rich nodes that are closer than Edge nodes to end‐users. Since these nodes might have enough resources to host services, execute tasks or even run containers, the utilization of network resources might be improved, and delay reduced by utilizing these nodes. The nodes must, therefore, be assessed to determine which nodes should offer resources to other nodes based on their situation. In this article, a new comparative assessment model for ranking Mist nodes in highly heterogeneous massive‐scale IoT networks in order to discover nodes that can offer their resources is proposed. The Mist nodes are evaluated based on parameters like resources, connections, applications, and environmental parameters to heuristically compare the neighbors with a novel learning‐to‐rank method to predict a suitability score for each node. The most suitable neighbor is then selected based on the score, with load balancing accomplished by a second chance method. When evaluating the performance, the results show that the proposed method succeeds in identifying resource‐rich nodes, while considering the selection of other nodes.
-
-
Tennebø, Frode & Geitle, Marius
(2019).
Evaluating Population Based Training on Small Datasets.
NIKT: Norsk IKT-konferanse for forskning og utdanning.
ISSN 1892-0713.
Full text in Research Archive
Show summary
Recently, there has been an increased interest in using artificial neural networks in the severely resource-constrained devices found in Internet-of-Things networks, in order to perform actions learned from the raw sensor data gathered by these devices. Unfortunately, training neural networks to achieve optimal prediction accuracy requires tuning multiple hyper-parameters, a process which has traditionally taken many times the computation time of a single training run of the neural network. In this paper, we empirically evaluate the Population Based Training algorithm, a method which simultaneously both trains and tunes a neural network, on datasets of similar size to what we might encounter in an IoT scenario. We determine that the population based training algorithm achieves prediction accuracy comparable to a traditional grid or random search on small datasets, and achieves state-of-the-art results for the Biodeg dataset.
-
Geitle, Marius & Olsson, Roland
(2019).
A New Baseline for Automated Hyper-Parameter Optimization.
In Nicosia, Giuseppe; Umeton, Renato; Sciacca, Vincenzo; Pardalos, Panos & Giuffrida, Giovanni (Ed.),
Machine Learning,Optimization,and Data Science - 5th International Conference, LOD 2019.
Springer Nature.
ISSN 9783030375980.
p. 521–530.
doi:
10.1007/978-3-030-37599-7_43.
Full text in Research Archive
-
-
View all works in NVA
Published June 12, 2018 4:14 PM
- Last modified Dec. 1, 2022 6:47 PM
