Wyniki wyszukiwania - BazTech

Ograniczanie wyników

Powiadomienia systemowe

Sesja wygasła!

Znaleziono wyników: 2

Liczba wyników na stronie

Wyniki wyszukiwania

Wyszukiwano:
w słowach kluczowych: RandomForest

Sortuj według:

Ogranicz wyniki do:

Tunneling Activities Detection Using Machine Learning Techniques

Allard F., Dubois R., Gompel P., Morel M.

Journal of Telecommunications and Information Technology

2011

nr 1

37-42

Tunnel establishment, like HTTPS tunnel or related ones, between a computer protected by a security gateway and a remote server located outside the protected network is the most effective way to bypass the network security policy. Indeed, a permitted protocol can be used to embed a forbidden one until the remote server. Therefore, if the resulting information flow is ciphered, security standard tools such as application level gateways (ALG), firewalls, intrusion detection system (IDS), do not detect this violation. In this paper, we describe a statistical analysis of ciphered flows that allows detection of the carried inner protocol. Regarding the deployed security policy, this technology could be added in security tools to detect forbidden protocols usages. In the defence domain, this technology could help preventing information leaks through side channels. At the end of this article, we present a tunnel detection tool architecture and the results obtained with our approach on a public database containing real data flows.

Randomforest based assessment of the hERG channel inhibition potential for the early drug cardiotoxicity testing

Wiśniowska B., Mendyk A., Polak M., Szlęk J., Polak S.

Bio-Algorithms and Med-Systems

2010

Vol. 6, no. 12

131--136

Acquired long QT syndrome (LQTS) can lead to fatal ventricular arrhythmia and one of the most common reasons for developing LQTS seen in clinical settings as Torsade de Pointes (TdP) are drugs. LQTS syndrome and TdP are principally caused by the inhibition of the potassium channels encoded by hERG (the human ether-a-go-go related gene). The potassium channels and ionic currents (lkr llo, lks and others) together with calcium and sodium channels and currents (lCaL, lNa respectively) are the key elements of the electrophysiological interplay in heart. Drugs affinity to hERG channels and life-threatening interferences in heart electrophysiology resulted in withdrawal of many substances from the pharmaceutical market and some other drugs were black-boxed as potentially dangerous. Aim of the study was to develop reliable and easy to use model for the drug affinity to the hERG channel inhibition. Database used for the modeling purposes contains 447 records which were utilized during the modeling and validation levels. Dataset is freely available from the CompTox project website (www.tox-portal. net). Three various validation modes were applied to the model performance assessment to ensure highest possible reliability of the finał model: standard 10-fold cross validation procedure (10-fold CV), enhanced 10-fold cross validation (whole drugs excluded from test sets) and validation on external test set of 62 records for both previously present (different in vitro models) and absent in native dataset drugs. Pre-processing included recalculation of the original output (IC50 value - concentration of a drug which causes 50% inhibition of the ionic current) derived from the in vitro experiments, with use of the scaling factors. Random Forest algorithm with either 10 or 50 or 100 generated trees and unlimited tree depth implemented in WEKA software was used. The input consisted of 1034 parameters describing in vitro setting (3), physico-chemical properties (7), and structure (so called Chemical fingerprint -1024). Output had a binary characteristic with IC50 equal to 1 ĘM concentration as the safety threshold value (encoded as 0-safe, 1- unsafe). The performance of the best model estimated in simple 10-fold CV was 85% (1-88%, 0-82%) with an average ROC accuracy of 0.92. Implementation of rigorous 10-fold CV procedurę resulted in decrease in total accuracy to 72% (1-72%, 0-72%) with ROC value equal to 0.791. Test on the external set consists of three measures: all 62 records (total - 73%, 1-62%, 0-81%), 33 enew f records describing previously unknown drugs (total - 73%, 1-62%, 0-81%) and eold f records describing previously present drugs (total - 83%, 1-78%, 091%).