Machine learning and data analysis: solving practical problems with the winners of the industrial hackathon LC

How to calculate the intentions of cybercriminals attacking an industrial object and recognize the weak SOS signals that the industrial control system periodically sends against the background of “normal” behavior - let's talk about this and many other things next Wednesday at the CoLab: Deep Learning in the central office of "Kaspersky Lab". All those who are not indifferent to the topic of industrial safety will enjoy an exciting immersion into the world of machine learning and data analysis under the guidance of the winners of the spring industrial hackathon LC and experts of our company.

The heroes of the evening will talk about how to fight with a huge 44 GB data set of technological signals of the industrial system and emerge victorious, finding attacks that are almost impossible to identify. Moreover, two solutions to this task will be presented to the public - from the participants of the hackathon and from its compilers.

So, the winners of the hackathon will talk about how:
')

• to turn the finding of anomalous behavior in the logs of factory devices into the task of ordinary classification;
• how in a short time to generate light, but useful features that characterize the entire time series;
• how to divide the work within the team;
• how “on the knee” to mix different models on one set of features;
• how to raise AUC using kNN - cherries on Competitive Data Science cake.

A small spoiler: the approach of our heroes is based on the "three pillars":

1. a key decision based on counting time series statistics as features;
2. coding consecutive sparse states of a system with a feature vector;
3. detection of anomalies using LSTM, ideas on their modifications.

Well, the experts from Kaspersky Lab will share their vision of the problem of analyzing the data of the ACU TP: what criteria did they follow when composing the tasks and the data set for the participants, what they expected from the teams and what happened in the end.

In short, on June 7 at 19.00 we are waiting for everyone to practice Deep Machine Learning at the headquarters of the LC. Register for the event here .