Academia.eduAcademia.edu

Figure 3 – uploaded by Tahseen Ullah Khan

See full PDFdownloadDownload figure
Fig. 3. Workflow of TSSAP. Because TSSAP has produced promising clustering results, we will primarily focus on this method in the following subsections to explain it in detail.

Figure 3 Workflow of TSSAP. Because TSSAP has produced promising clustering results, we will primarily focus on this method in the following subsections to explain it in detail.

Related Figures (15)

A comparison on resource parameters and algorithms with related work. Table 1
A comparison on resource parameters and algorithms with related work. Table 1
A comparison on power model and algorithms with related work. Table 2
A comparison on power model and algorithms with related work. Table 2
Fig. 1. fastStorage: CPU Usage [%] sample at 5 min interval.
Fig. 1. fastStorage: CPU Usage [%] sample at 5 min interval.
Fig. 2. Proposed system model.
Fig. 2. Proposed system model.
Distributed cloud data centre’s trace for this work. Table 3 Table 4
Distributed cloud data centre’s trace for this work. Table 3 Table 4
Definition of features in workload trace.
Definition of features in workload trace.
Finally, using this similarity matrix as shown in Eq. (13), a fine set of potential exemplars is obtained. We also limited AP to producing a random number of exemplars by using a small amount of supervised data K that is passed into AP’s input. The accuracy of clustering improved as a result of this. The TSSAP pseudo code is shown in Algorithm 1:
Finally, using this similarity matrix as shown in Eq. (13), a fine set of potential exemplars is obtained. We also limited AP to producing a random number of exemplars by using a small amount of supervised data K that is passed into AP’s input. The accuracy of clustering improved as a result of this. The TSSAP pseudo code is shown in Algorithm 1:
GRU model training at different hyper parameters.
GRU model training at different hyper parameters.
Fig. 4. Forecasting results with GRU for different features of workloads in fastStorage traces. Fig. 5. Forecasting results with GRU for different features of workloads in Rnd traces.
Fig. 4. Forecasting results with GRU for different features of workloads in fastStorage traces. Fig. 5. Forecasting results with GRU for different features of workloads in Rnd traces.
Fig. 7. Rnd: GRU-Model train vs Validation Loss.
Fig. 7. Rnd: GRU-Model train vs Validation Loss.
Fig. 6. fastStorage: GRU-Model train vs. validation loss.
Fig. 6. fastStorage: GRU-Model train vs. validation loss.
Fig. 8. Clustering accuracy comparison for E-state evaluated using micro. precision.
Fig. 8. Clustering accuracy comparison for E-state evaluated using micro. precision.
Related topics:
Information SystemsComputer ScienceDistributed Computing
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved