An experimental evaluation of extreme learning machines on several hardware devices
- PDF / 1,917,377 Bytes
- 13 Pages / 595.276 x 790.866 pts Page_size
- 4 Downloads / 203 Views
(0123456789().,-volV)(0123456789(). ,- volV)
EXTREME LEARNING MACHINE AND DEEP LEARNING NETWORKS
An experimental evaluation of extreme learning machines on several hardware devices Liang Li1 • Guoren Wang2 • Gang Wu1,3 • Qi Zhang2 Received: 31 December 2018 / Accepted: 29 August 2019 Ó Springer-Verlag London Ltd., part of Springer Nature 2019
Abstract As an important learning algorithm, extreme learning machine (ELM) is known for its excellent learning speed. With the expansion of ELM’s applications in the field of classification and regression, the need for its real-time performance is increasing. Although the use of hardware acceleration is an obvious solution, how to select the appropriate acceleration hardware for ELM-based applications is a topic worthy of further discussion. For this purpose, we designed and evaluated the optimized ELM algorithms on three kinds of state-of-the-art acceleration hardware, i.e., multi-core CPU, Graphics Processing Unit (GPU), and Field-Programmable Gate Array (FPGA) which are all suitable for matrix multiplication optimization. The experimental results showed that the speedup ratio of these optimized algorithms on acceleration hardware achieved 10–800. Therefore, we suggest that (1) use GPU to accelerate ELM algorithms for large dataset, and (2) use FPGA for small dataset because of its lower power, especially for some embedded applications. We also opened our source code. Keywords Extreme learning machine Hardware Multi-core GPU FPGA
1 Introduction As machine learning technologies (e.g., Support Vector Machine [35], Neural Networks [6, 26], and Random Forest [21]), especially deep learning [2, 3, 27], are applied in a continuously wider range of scenarios, people are paying more and more attention to the performance of these algorithms. However, the learning speed of traditional machine learning algorithms is widely criticized & Guoren Wang [email protected] Liang Li [email protected] Gang Wu [email protected] Qi Zhang [email protected] 1
Computer Science and Engineering, Northeastern University, Shenyang, China
2
Computer Science and Technology, Beijing Institute of Technology, Beijing, China
3
State Key Laboratory for Novel Software Technology, Nanjing University, Nanjing 210023, China
[14]. The main reason for the slow speed is that parameters of machine learning algorithms usually need to be updated iteratively in a gradient method. Hence, traditional machine learning methods are difficult to meet the real-time learning needs for large-scale data applications. Extreme learning machine (ELM) [9–12, 14–16] is a feedforward neural network whose design objective is to ensure a high accuracy, least user intervention, and real-time learning [14]. In practical applications, ELM is usually superior to the traditional machine learning algorithms, such as SVM and back propagation (BP) in both classification accuracy and learning speed when fed with adequate training samples [11]. Therefore, ELM has found application scenarios in medical healthcare
Data Loading...
