Highlights - Computational Intelligence
Principal Investigator: Simone Ludwig (Computer Science, North Dakota State University)
Dr. Ludwig’s research focuses on computational intelligence which is part of artificial intelligence and is concerned with evolutionary computation, swarm intelligence, deep neural networks, and fuzzy logic and reasoning.
An emphasis is on the parallelization of nature-inspired algorithms using frameworks such as Spark to speed up the execution time of those stochastic population-based algorithms [1–4]. For example, a scalable design and implementation of a particle swarm optimization classification (SCPSO) approach that is based on the Apache Spark framework was proposed [2]. The main idea of the SCPSO algorithm is to find the optimal centroid for each target label using particle swarm optimization and then assign unlabeled data points to the closest centroid. Two versions of SCPSO, SCPSO-F1 and SCPSO-F2, were proposed based on different fitness functions, which were tested on real datasets to evaluate their scalability and performance. The results revealed that SCPSO-F1 and SCPSO-F2 scale very well with increasing dataset sizes and the speedup of SCPSO-F2 is almost identical to the optimal speedup, while the speedup of SCPSO-F1 is very close to the optimal speedup. Consequently, they can be efficiently parallelized using the Apache Spark framework which runs on cluster nodes.
The architecture of the implementation and the results of speedup and scaleup are shown in the figure below.
Another research area being pursued by Dr. Ludwig’s group is in the area of machine learning where big data is involved, in particular, applying classification to areas in cybersecurity [5,6], medicine [7] and cloud computing [8,9].
Both areas require infrastructure for the large data storage and heavy computing needs. Graphics processing units (GPUs) have often been used to run TensorFlow, Keras, and PyTorch. In addition to the HPC resources provided by CCAST, the group has also used national computing facilities provided via the NSF-sponsored Extreme Science and Engineering Discovery Environment (XSEDE).
References
[1] J. Al-Sawwa and S. A. Ludwig, “Parallel Particle Swarm Optimization Classification Algorithm Variant implemented with Apache Spark,” Concurrency Computat. Pract. Exper. 32, e5451 (2020).
[2] G. Miryala, S. A. Ludwig, “Comparing Spark with MapReduce - Glowworm Swarm Optimization applied to Multimodal Functions,” Int. J. Swarm. Intell. Res. 9, 1 (2018).
[3] I. Aljarah, S. A. Ludwig, “A Scalable MapReduce-enabled Glowworm Swarm Optimization Approach for High Dimensional Multimodal Functions,” Int. J. Swarm. Intell. Res. 7, 34 (2016).
[4] S. A. Ludwig, “MapReduce-based Fuzzy C-Means Clustering Algorithm: Implementation and Scalability,” Int. J. Mach. Learn. Cyb. 6, 923 (2015).
[5] S. A. Ludwig, “Applying a Neural Network Ensemble to Intrusion Detection,” J. Artif. Intell. Soft Comput. Res. 9, 177 (2019).
[6] S. A. Ludwig, “Intrusion Detection of Multiple Attack Classes using a Deep Neural Net Ensemble,” IEEE Symposium Series on Computational Intelligence (SSCI), Honolulu, HI, USA, Oct. 2017
[7] M. F. Kabir and S. A. Ludwig, “Enhancing the Performance of Classification Using Super Learning,” Data-Enabled Discov. Appl. 3, 5 (2019).
[8] S. A. Ludwig, K. Bauer, “Immune Network Algorithm applied to the Optimization of Composite SaaS in Cloud Computing,” 2015 IEEE Congress on Evolutionary Computation, Sendai, Japan.
[9] S. A. Ludwig, “Particle Swarm Optimization Approach with Parameter-wise Hill-climbing Heuristic for Task Allocation of Workflow Applications on the Cloud,” Proceedings of IEEE International Conference on Tools with Artificial Intelligence (ICTAI), Washington DC, USA.