Difference between revisions of "IoTSec:Phone 2020/02/12"
From its-wiki.no
(→Status) |
(→Status) |
||
(2 intermediate revisions by the same user not shown) | |||
Line 55: | Line 55: | ||
* | * | ||
− | NR | + | NR has studied and tested predictive analytics based on machine learning algorithms. At this stage, the following machine learning algorithms have been selected for the predictive analytics toolkit: Support Vector Machine (SVM), K- nearest neighbors (KNN), Decision Tree, Random Forest. The toolkit has been tested using the following three different datasets. |
− | has studied and tested predictive analytics based on machine learning algorithms. At this stage, the following machine learning algorithms have been selected for the predictive analytics toolkit: Support Vector Machine (SVM), K- nearest neighbors (KNN), Decision Tree, Random Forest. The toolkit has been tested using the following three different datasets. | + | |
The KDDCup-99 dataset [http://kdd.ics.uci.edu/databases/kddcup99/kddcup99.html]. | The KDDCup-99 dataset [http://kdd.ics.uci.edu/databases/kddcup99/kddcup99.html]. | ||
The CIC IDS 2017 dataset [https://www.unb.ca/cic/datasets/ids-2017.html]. | The CIC IDS 2017 dataset [https://www.unb.ca/cic/datasets/ids-2017.html]. | ||
Line 62: | Line 61: | ||
This preliminary study has shown that the SVM method has performed inadequately for training /testing time. | This preliminary study has shown that the SVM method has performed inadequately for training /testing time. | ||
It has also achieved lower accuracy for the UNSW-NB15 Dataset. The random forest method performs well while requiring slightly more time for training than the decision tree method. In this preliminary study, we have used the datasets that available on line. All features supplied with these datasets have been applied. OpenAPI is also implemented using swagger. | It has also achieved lower accuracy for the UNSW-NB15 Dataset. The random forest method performs well while requiring slightly more time for training than the decision tree method. In this preliminary study, we have used the datasets that available on line. All features supplied with these datasets have been applied. OpenAPI is also implemented using swagger. | ||
− | + | Next, we plan to (1) define how to select a feature set that produces acceptable results with predefined accuracy while reducing the volume of the collected and stored data, (2) develop methods for predictive analytics that operate on real-time data collections and investigate new efficient predictive algorithms based on deep learning techniques, and (3) contribute to the paper based on the results from (1) and (2). | |
== other discussion == | == other discussion == | ||
{{IoTSec:ActionItems}} | {{IoTSec:ActionItems}} |
Latest revision as of 14:20, 12 February 2020
Security in IoT for Smart Grids | |||||||
---|---|---|---|---|---|---|---|
|
![]() |
Title | IoTSec:Phone Meeting February2020 |
---|---|
Phone number, info | +47 21933751 725-796-213 |
Date, time, Duration | 2020/02/12, 14:00, 60 min |
Organizer | Christian Johansen |
Participants | Christian Johansen |
this page was created by Special:FormEdit/Phone_IoTSec, and can be edited by Special:FormEdit/Phone_IoTSec/IoTSec:Phone 2020/02/12
Call Info
http://gotomeet.me/BasicInternet
or phone Norway: +47 21 93 37 51 Access Code: 725-796-213
More phone numbers FI: +358 923 17 0568 DE: +49 692 5736 7317 IT: +39 0 247 92 13 01 NL: +31 202 251 017 ES: +34 932 75 2004 SE: +46 853 527 827
Help on joining can be taken from IoTSec:Meetings#Help_for_using_GoToMeeting_conference_tool
Agenda
- Status for the joint IoTSec papers and general project goal fitting
- Progress plans with Milestones and Responsibilities
- Status from each partner (at least from PhD students including update on time-line)
- AOB
Minutes
Minutes of the meeting IoTAdmin:20200212PhoneMeeting_Notes
short notes
- What are the Main Takeaways from IoTSec ?
- What are the Major achievements?
- Link to common paper https://docs.google.com/document/d/1Gj4nZ3YkUUaByEsDruhx5LqyWt_G-M_2czMxM4lRtVI/edit
- One-pager ? for IoTSec
- Maunya from SCOTTo to talk about the hackaton with SGSC
Status
UiO
SIN
NTNU
NR has studied and tested predictive analytics based on machine learning algorithms. At this stage, the following machine learning algorithms have been selected for the predictive analytics toolkit: Support Vector Machine (SVM), K- nearest neighbors (KNN), Decision Tree, Random Forest. The toolkit has been tested using the following three different datasets. The KDDCup-99 dataset [1]. The CIC IDS 2017 dataset [2]. The UNSW-NB15 dataset [3]. This preliminary study has shown that the SVM method has performed inadequately for training /testing time. It has also achieved lower accuracy for the UNSW-NB15 Dataset. The random forest method performs well while requiring slightly more time for training than the decision tree method. In this preliminary study, we have used the datasets that available on line. All features supplied with these datasets have been applied. OpenAPI is also implemented using swagger. Next, we plan to (1) define how to select a feature set that produces acceptable results with predefined accuracy while reducing the volume of the collected and stored data, (2) develop methods for predictive analytics that operate on real-time data collections and investigate new efficient predictive algorithms based on deep learning techniques, and (3) contribute to the paper based on the results from (1) and (2).
other discussion