WEBVTT

1
00:00:13.666 --> 00:00:14.226
I'm Sutanay.

2
00:00:14.416 --> 00:00:15.886
I'm a senior research scientist

3
00:00:15.886 --> 00:00:17.006
at Pacific Northwest

4
00:00:17.006 --> 00:00:18.236
National Laboratory.

5
00:00:18.546 --> 00:00:20.236
We have developed a technology called

6
00:00:20.236 --> 00:00:22.376
StreamWorks, which can find emerging

7
00:00:22.376 --> 00:00:24.436
patterns of behavior in streaming data.

8
00:00:25.256 --> 00:00:26.276
So the problem we are trying

9
00:00:26.276 --> 00:00:28.576
to solve right now is that most

10
00:00:28.576 --> 00:00:29.966
of cyber security works today

11
00:00:29.966 --> 00:00:31.146
in a very reactive fashion.

12
00:00:31.416 --> 00:00:32.256
Something happens,

13
00:00:32.746 --> 00:00:34.486
in the aftermath we try to dig back

14
00:00:34.486 --> 00:00:36.586
into the data and see what happened.

15
00:00:36.926 --> 00:00:38.936
And what you're trying to change is turn

16
00:00:38.936 --> 00:00:40.576
that reactive mode of operation

17
00:00:40.576 --> 00:00:41.696
into a more proactive mode.

18
00:00:42.086 --> 00:00:45.466
We look for patterns of behavior instead

19
00:00:45.466 --> 00:00:46.566
of looking for exact

20
00:00:46.736 --> 00:00:48.226
or all possible ways an attack

21
00:00:48.226 --> 00:00:48.726
can happen.

22
00:00:49.226 --> 00:00:50.936
Every smart attacker tries

23
00:00:50.936 --> 00:00:52.506
to hide their identity,

24
00:00:53.376 --> 00:00:55.656
but even for them changing the patterns

25
00:00:55.656 --> 00:00:56.846
of behavior is difficult.

26
00:00:57.816 --> 00:00:58.896
Our technology comes

27
00:00:58.896 --> 00:01:00.966
with three major benefits, novelty,

28
00:01:01.566 --> 00:01:02.816
that we are trying

29
00:01:02.816 --> 00:01:05.096
to find behavioral patterns instead

30
00:01:05.096 --> 00:01:07.166
of looking for all possible ways an

31
00:01:07.166 --> 00:01:09.676
attack can emerge; efficiency --

32
00:01:10.076 --> 00:01:12.046
we have developed new incremental

33
00:01:12.216 --> 00:01:14.346
algorithms that can only look

34
00:01:14.346 --> 00:01:15.856
at the newest information

35
00:01:16.146 --> 00:01:17.616
and does minimal processing

36
00:01:18.006 --> 00:01:18.926
to be efficient.

37
00:01:19.556 --> 00:01:21.386
We take advantage of parallel computing

38
00:01:21.716 --> 00:01:23.916
to even further achieve scaling.

39
00:01:24.376 --> 00:01:25.716
And third is usability.

40
00:01:27.406 --> 00:01:28.756
We address a major gap

41
00:01:28.756 --> 00:01:29.626
in the market today.

42
00:01:29.836 --> 00:01:31.076
There are streaming technologies

43
00:01:31.156 --> 00:01:31.716
and there are graph

44
00:01:31.716 --> 00:01:32.626
database technologies.

45
00:01:32.796 --> 00:01:35.096
But nobody can perform complex graph

46
00:01:35.096 --> 00:01:36.596
pattern matching in streaming fashion.

47
00:01:36.886 --> 00:01:38.266
That's what StreamWorks does.

48
00:01:38.756 --> 00:01:39.776
Stopping cyber-attacks

49
00:01:39.776 --> 00:01:42.016
in their tracks requires finding complex

50
00:01:42.016 --> 00:01:43.946
patterns of behavior in near real-time

51
00:01:44.326 --> 00:01:45.816
and StreamWorks will do that for you.