WEBVTT 1 00:00:00.000 --> 00:00:04.010 [Host] Hello and welcome to today's Tech Talk with the department of Homeland Security 2 00:00:04.010 --> 00:00:08.080 Science and Technology Directorate. This is the first part 3 00:00:08.080 --> 00:00:12.090 of our brand new, live video interactive series. But we're gonna highlight some of S&T's 4 00:00:12.090 --> 00:00:16.180 projects and get an opportunity to talk to one of the Program Managers in real-time. 5 00:00:16.180 --> 00:00:20.210 So today, I am joined by Program Manager Darren Wilson. He's gonna talk to us about 6 00:00:20.210 --> 00:00:24.230 Screen Adapt and Identify, which are two training capabilities that we have here at S&T. 7 00:00:24.230 --> 00:00:28.260 But first, I want to kick things off by watching a quick video. 8 00:00:28.260 --> 00:00:32.330 [Video Music playing] 9 00:00:32.330 --> 00:00:37.330 Screeners Analyze thousands of images daily. 10 00:00:40.350 --> 00:00:44.450 Anomalies may be unlikely, but the risk of a miss is high. 11 00:00:44.450 --> 00:00:48.480 Screeners must become visual search experts by being 12 00:00:48.480 --> 00:00:52.520 both effective and efficient. Screen Adapt provides a 13 00:00:52.520 --> 00:00:56.610 personalized path to visual search expertise. Develop 14 00:00:56.610 --> 00:01:00.760 core visual search skills faster through an individualized training path 15 00:01:00.760 --> 00:01:04.820 that addresses skill deficiencies. Immediate feedback that keeps the trainee aware 16 00:01:04.820 --> 00:01:08.920 of their progress and errors. Dynamically created combinations of 17 00:01:08.920 --> 00:01:12.930 threat and bag images that keeps training fresh and unpredictable. 18 00:01:12.930 --> 00:01:16.940 Screen Adapt uses a personalized training approach 19 00:01:16.940 --> 00:01:20.960 based on visual search behavior and performance. 20 00:01:20.960 --> 00:01:25.020 Eye tracking makes the visual search process observable to provide immediate 21 00:01:25.020 --> 00:01:29.050 feedback. Eye tracking makes it possible to determine the root cause of visual 22 00:01:29.050 --> 00:01:33.070 search errors. Training content is then changed 23 00:01:33.070 --> 00:01:37.150 in real-time to address the identified skill deficiencies. 24 00:01:37.150 --> 00:01:41.290 Screen adapt also allows instructors to review and manage the training process. 25 00:01:41.290 --> 00:01:45.330 Upload new images to keep training relevant and up to 26 00:01:45.330 --> 00:01:49.410 date and analyze training progress with training management tools. 27 00:01:49.410 --> 00:01:53.530 Screen adapt. [To learn more about ScreenADAPT, please contact Darren P. Wilson - Program Manager 28 00:01:53.530 --> 00:01:57.530 Science and Technology Directorate Darren.Wilson@hq.dhs.gov] 29 00:01:57.530 --> 00:02:01.540 [Video and Music Ends] 30 00:02:01.540 --> 00:02:05.590 Instead of just a high level overview of what screen adapt 31 00:02:05.590 --> 00:02:09.660 can do. What I want to start by doing is just asking a few 32 00:02:09.660 --> 00:02:13.670 questions then we can open it up to the audience. [Clears Throat] Excuse me. So Darren can you give 33 00:02:13.670 --> 00:02:17.780 us an idea of what these technologies do and why they were developed? 34 00:02:17.780 --> 00:02:21.870 [Darren] Absolutely, there's cross coming means 35 00:02:21.870 --> 00:02:25.890 across the Homeland Security Enterprise and the law enforcement community 36 00:02:25.890 --> 00:02:29.950 that have to do with visual search. Screen Adapt 37 00:02:29.950 --> 00:02:34.050 is a new and innovative capability that leverages 38 00:02:34.050 --> 00:02:38.070 the latest in training research and the latest in 39 00:02:38.070 --> 00:02:42.130 eye tracking technology to maximize visual search skills. 40 00:02:42.130 --> 00:02:46.240 It's core capabilities at this point are 41 00:02:46.240 --> 00:02:50.270 advance x-ray image and masses training as well as 42 00:02:50.270 --> 00:02:54.340 imposter detection training. [Host] Okay, and what organizations have developed these technologies? 43 00:02:54.340 --> 00:02:58.450 Because S&T is really the research and development 44 00:02:58.450 --> 00:03:02.490 directorate of the Homeland Security Department. 45 00:03:02.490 --> 00:03:06.500 So who do you work with on these tehnologies? 46 00:03:06.500 --> 00:03:10.510 [Darren] These projects were done in collaboration with 47 00:03:10.510 --> 00:03:14.550 TSA, the office of training development, 48 00:03:14.550 --> 00:03:18.610 the screeners at the individual airports, 49 00:03:18.610 --> 00:03:22.630 federal security directors that were counting on to us 50 00:03:22.630 --> 00:03:26.670 to come in and work with screeners. Give us access to 51 00:03:26.670 --> 00:03:30.750 images and identify what was done in collaboration with Customs and Border Protection. 52 00:03:30.750 --> 00:03:34.770 [Host] So we went in and it sounds like we asked what were the needs that they were seeing 53 00:03:34.770 --> 00:03:38.810 as far as training capabilities and these are really 54 00:03:38.810 --> 00:03:42.920 specific to that eye tracking need that we saw. Can you talk more about the 55 00:03:42.920 --> 00:03:46.940 eye tracking capability? [Darren] Absolutely, so the eye tracking capability is 56 00:03:46.940 --> 00:03:50.990 really key, because it gives us information that we really didn't have 57 00:03:50.990 --> 00:03:55.140 before about screener performance. We can see 58 00:03:55.140 --> 00:03:59.150 where the screener is looking. What scan pattern their using. 59 00:03:59.150 --> 00:04:03.210 What their fixation times are and if they are applying a 60 00:04:03.210 --> 00:04:07.310 scan pattern consistently and completely. More importantly 61 00:04:07.310 --> 00:04:11.350 we can identify the root causes of errors. Meaning 62 00:04:11.350 --> 00:04:15.420 of they miss a treat we can time out if they missed the 63 00:04:15.420 --> 00:04:19.550 threat because they are making a scanning error or a recognition error. 64 00:04:19.550 --> 00:04:23.600 A scanning error is when the screener looks at the image 65 00:04:23.600 --> 00:04:27.680 but doesn't look at a particular area and that's why they 66 00:04:27.680 --> 00:04:31.800 miss a threat. A recognition error is when they search 67 00:04:31.800 --> 00:04:35.870 the image, look at a particular area where there's a threat and don't recognize 68 00:04:35.870 --> 00:04:39.970 that it is a threat. That's important because depending on what type of 69 00:04:39.970 --> 00:04:43.990 errors they are making, it requires a different type of training. If they are making 70 00:04:43.990 --> 00:04:48.050 a scanning error, the system adapts to give them more of 71 00:04:48.050 --> 00:04:52.140 your traditional exposure training. They can work on with 72 00:04:52.140 --> 00:04:56.170 the trainer and applying that their scan pattern more consistently and 73 00:04:56.170 --> 00:05:00.220 completely to make sure that their scan pattern covers the 74 00:05:00.220 --> 00:05:04.230 entire image. If they are making a recognition error that takes a different type of training 75 00:05:04.230 --> 00:05:08.260 which is a discrimination training which builds their mental 76 00:05:08.260 --> 00:05:12.330 threat image library so that they can better recognize threat items and non threat items 77 00:05:12.330 --> 00:05:16.440 moving forward. [Host] Excellent. So it's the difference between not seeing something 78 00:05:16.440 --> 00:05:20.480 in let's say a persons bag versus not recognizing that oh that may be a blade 79 00:05:20.480 --> 00:05:24.550 or that might be something that we definitely want to flag for an additional search. [Darren] Absolutely. 80 00:05:24.550 --> 00:05:28.650 [Host] So before this technology was available, what was the training capability? 81 00:05:28.650 --> 00:05:32.660 You know how is this meeting that demand for this new and innovative need? 82 00:05:32.660 --> 00:05:36.670 [Darren] Well, I think that you know obviously 83 00:05:36.670 --> 00:05:40.680 TSA does a great job at what they do, but we're trying to 84 00:05:40.680 --> 00:05:44.690 put more tools in their toolbox. Give them more diagnostic 85 00:05:44.690 --> 00:05:48.770 capabilities and leverage emerging technology. 86 00:05:48.770 --> 00:05:52.780 Right not they do exposure training, which is looking at the image to find 87 00:05:52.780 --> 00:05:56.830 the treats and the simulators 88 00:05:56.830 --> 00:06:00.840 actually based on the actual check point systems that 89 00:06:00.840 --> 00:06:04.840 they use. This system screen adapt is technology 90 00:06:04.840 --> 00:06:08.890 diagnostic, because we focusing on the core visual search 91 00:06:08.890 --> 00:06:12.980 skills and trying to improve those skills. So matter where 92 00:06:12.980 --> 00:06:17.000 the technology goes, which type of machine that their using 93 00:06:17.000 --> 00:06:21.060 it is impacting performance in a positive way, regardless 94 00:06:21.060 --> 00:06:25.070 of technology. [Host] So this training capability can really be used 95 00:06:25.070 --> 00:06:29.090 no matter what type of screening technology comes down the line someday. [Clears throat] Excuse me. 96 00:06:29.090 --> 00:06:33.140 I want to go back to something you talked about. The 97 00:06:33.140 --> 00:06:37.240 different types of errors and how the instructor knows what errors somebody is making. 98 00:06:37.240 --> 00:06:41.270 That's hard to judge because so much of the screening is based on whats happening in somebodies head. 99 00:06:41.270 --> 00:06:45.330 [Darren] Absolutely. [Host] So what type of information then does screen adapt then identify 100 00:06:45.330 --> 00:06:49.430 provide the instructor, you know as far as how this trainee is performing? 101 00:06:49.430 --> 00:06:53.460 [Darren] It gives, the eye tracker is able to 102 00:06:53.460 --> 00:06:57.530 focus in on and then display back to the trainer and trainee 103 00:06:57.530 --> 00:07:01.530 where they looked, how long they looked and what sequence they looked. 104 00:07:01.530 --> 00:07:05.570 And that information coupled with the types of threats that 105 00:07:05.570 --> 00:07:09.650 their missing. The level of difficulty of threats that they are missing are all 106 00:07:09.650 --> 00:07:13.760 going to the adaptive nature of the system. If they are missing more 107 00:07:13.760 --> 00:07:17.810 guns, then they are going to see more guns. If their missing more knives then they are going to see 108 00:07:17.810 --> 00:07:21.900 more knives. ID's, ID components, other prohibited items. 109 00:07:21.900 --> 00:07:25.910 It's pulling from a database where we scanned 110 00:07:25.910 --> 00:07:29.960 the badge and the threats independently so we can have 111 00:07:29.960 --> 00:07:34.050 infinite number of combinations of threats and banks to customize 112 00:07:34.050 --> 00:07:38.070 to that individual trainees specific needs. [Host] Wow. 113 00:07:38.070 --> 00:07:42.150 So they can do that almost instantaneously as somebody is taking this training and say you know your 114 00:07:42.150 --> 00:07:46.240 missing a couple here, so I'm going to throw you some more so you get that better practice. 115 00:07:46.240 --> 00:07:50.260 What was it like to try to develop that capability? [Darren] It's 116 00:07:50.260 --> 00:07:54.320 definitely a group effort. We collaborated with all the 117 00:07:54.320 --> 00:07:58.420 different groups within TSA to make it happen. 118 00:07:58.420 --> 00:08:02.450 It's important to them to stay 119 00:08:02.450 --> 00:08:06.510 on the cutting edge of technology and be testing up 120 00:08:06.510 --> 00:08:10.520 new and innovative solutions. We have a situation where 121 00:08:10.520 --> 00:08:14.550 did a [inteeligible] test that screened it out specifically 122 00:08:14.550 --> 00:08:18.620 for x-ray image analysis training. Actually resulted in 123 00:08:18.620 --> 00:08:22.730 screeners being able to clear bags faster, 124 00:08:22.730 --> 00:08:26.770 clear bags with no threats faster, which 125 00:08:26.770 --> 00:08:26.810 operationally would correlate to faster 126 00:08:26.810 --> 00:08:30.890 operationally would correlate to faster [unintelligible]. 127 00:08:30.890 --> 00:08:34.910 Things that you would actually see at the check point in terms of efficiency. 128 00:08:34.910 --> 00:08:38.950 The first fixation on actual threats was faster. 129 00:08:38.950 --> 00:08:43.030 And the identification of threats was also faster, 130 00:08:43.030 --> 00:08:47.040 without anyway of adversely effecting the effectiveness 131 00:08:47.040 --> 00:08:51.090 of the screeners finding the the threats which is really critical. [Host] Absolutely, because we all just 132 00:08:51.090 --> 00:08:55.190 want to get through the line faster. We want to make sure we're getting through the line safely. 133 00:08:55.190 --> 00:08:59.200 When folks when they get to the airport they know that baggage being checked correctly 134 00:08:59.200 --> 00:09:03.250 and accurately. So it sounds likes this is a great way to both speed up the line 135 00:09:03.250 --> 00:09:07.350 for the lack of a better time, but also make sure that it's really well 136 00:09:07.350 --> 00:09:11.370 while effectively screening for those things. So we have a quick question from 137 00:09:11.370 --> 00:09:15.430 facebook. When will screen adapt and identify be used at airports? 138 00:09:15.430 --> 00:09:19.530 [Darren] Well right now, we have systems deployed 139 00:09:19.530 --> 00:09:23.540 at airports. About 50 systems in the training 140 00:09:23.540 --> 00:09:27.610 effectiveness evaluation stage. We're collecting qualitative 141 00:09:27.610 --> 00:09:31.720 and quantitative data on that larger training effectiveness 142 00:09:31.720 --> 00:09:35.800 evaluation to be able to draw some broader conclusions 143 00:09:35.800 --> 00:09:39.830 and extrapolate from the pilots that we've 144 00:09:39.830 --> 00:09:43.940 did. Originally we tested the system with 115 145 00:09:43.940 --> 00:09:47.990 screeners at 3 different airports that resulted in a 45% 146 00:09:47.990 --> 00:09:52.070 increase in efficiency. So we're 147 00:09:52.070 --> 00:09:56.080 really excited about the initial results, we're really excited about the expanded 148 00:09:56.080 --> 00:10:00.120 training effectiveness evaluation at 149 00:10:00.120 --> 00:10:04.210 six airports that TSA is currently conducting 150 00:10:04.210 --> 00:10:08.230 and looking forward to the results that are coming back from that. 151 00:10:08.230 --> 00:10:12.290 [Host] So in addition to that data both qualitative and quantitative, what kind of feedback are you getting from 152 00:10:12.290 --> 00:10:16.390 people who have used this training? [Darren] The feedback 153 00:10:16.390 --> 00:10:20.410 has been super positive. We have airports that are participating in the training effectiveness evaluation 154 00:10:20.410 --> 00:10:24.470 that actually have requested to keep the systems, use them 155 00:10:24.470 --> 00:10:28.560 as part of their training program. We've had 156 00:10:28.560 --> 00:10:32.780 CBP actually after the training effecting 157 00:10:32.780 --> 00:10:36.790 this with identifying actually 158 00:10:36.790 --> 00:10:40.890 roll those identified systems right into training for our 159 00:10:40.890 --> 00:10:44.930 new CBP officers. [Host] Oh wow. [Darren] So CBP has 160 00:10:44.930 --> 00:10:49.000 is going to have 12 systems total that they are using at their 161 00:10:49.000 --> 00:10:53.010 training academy down at the enforcement training center to train all 162 00:10:53.010 --> 00:10:57.050 new CBP officers in the imposter identification task. 163 00:10:57.050 --> 00:11:01.130 [Host] That's great. We have another question that we want to respond to. 164 00:11:01.130 --> 00:11:05.140 What are the main motivations that led to this effort. 165 00:11:05.140 --> 00:11:09.190 [Darren] I think they were always seeking to improve 166 00:11:09.190 --> 00:11:13.270 the efficiency and effectiveness of 167 00:11:13.270 --> 00:11:17.290 our screening checkpoints, but really improve the passenger experience 168 00:11:17.290 --> 00:11:21.340 as well. If we can eliminate unnecessary secondary 169 00:11:21.340 --> 00:11:25.430 searches that everybody experiences every once in a while 170 00:11:25.430 --> 00:11:29.450 when their going through the airports, that is a huge win. 171 00:11:29.450 --> 00:11:33.460 The more people that we can screen 172 00:11:33.460 --> 00:11:37.550 the less the wait time is and the more efficient we 173 00:11:37.550 --> 00:11:41.570 are at checkpoints. [Host] So this really benefits both the operators and the 174 00:11:41.570 --> 00:11:45.630 american public in terms of more effective screening, 175 00:11:45.630 --> 00:11:49.740 getting through the checkpoints faster, but also knowing that you are still safe getting through there? 176 00:11:49.740 --> 00:11:53.770 [Darren] Absolutely! It also save those training organizations time 177 00:11:53.770 --> 00:11:57.840 because you're able to bring new hires up to full 178 00:11:57.840 --> 00:12:01.950 performance level faster and give the trainers 179 00:12:01.950 --> 00:12:05.980 new diagnostic tools. We have after action review 180 00:12:05.980 --> 00:12:10.100 screening gives information like, how did this 181 00:12:10.100 --> 00:12:14.110 individual do this compared to their peers? How did they do as compared to 182 00:12:14.110 --> 00:12:18.150 how they thought they did? How are they progressing through the training? What is their breakdown of 183 00:12:18.150 --> 00:12:22.240 hits, misses, false alarms to really 184 00:12:22.240 --> 00:12:26.250 tailor their training experience to meet their individual needs. 185 00:12:26.250 --> 00:12:30.300 [Host] Thats great. We have another question. Will this technology be used for 186 00:12:30.300 --> 00:12:34.380 large scale NII x-ray that ports are borders? 187 00:12:34.380 --> 00:12:38.400 [Darren] We're exploring other options with 188 00:12:38.400 --> 00:12:42.450 CBP. They do all sorts of 189 00:12:42.450 --> 00:12:46.450 different screening when it comes to agriculture, 190 00:12:46.450 --> 00:12:50.470 vehicles. We are looking at 191 00:12:50.470 --> 00:12:54.520 applications beyond just x-ray image analysis and 192 00:12:54.520 --> 00:12:58.610 the impostor detection 193 00:12:58.610 --> 00:13:02.640 and looking to branching it out into other visual search 194 00:13:02.640 --> 00:13:06.700 tasks like, suspicious behavior detection and pre-deciding indicators for 195 00:13:06.700 --> 00:13:10.710 our law enforcement community. [Host] That's great! Another question just came in. 196 00:13:10.710 --> 00:13:14.740 Do you see this as something to be used in daily operations in the field, 197 00:13:14.740 --> 00:13:18.800 rather than only using this as part of training? 198 00:13:18.800 --> 00:13:22.900 We've got a second part of the question 199 00:13:22.900 --> 00:13:26.930 coming in. Check point environments is totally different 200 00:13:26.930 --> 00:13:30.930 than classroom. [Darren] Absolutely and there is 201 00:13:30.930 --> 00:13:35.050 all sorts of different other variables that go into 202 00:13:35.050 --> 00:13:39.060 the checkpoint screening environment. Whether your talking about a federal building 203 00:13:39.060 --> 00:13:43.130 people going into a federal buiding. Whether your talking 204 00:13:43.130 --> 00:13:47.150 about CBP checkpoint or a 205 00:13:47.150 --> 00:13:51.190 TSA checkpoint. There are lots of different variables that 206 00:13:51.190 --> 00:13:55.280 go into all those things. In addition to the non biology 207 00:13:55.280 --> 00:13:59.290 of the different technologies that are used. 208 00:13:59.290 --> 00:14:03.340 The basic visual search skills and try to maximize the 209 00:14:03.340 --> 00:14:07.420 efficiency and effectiveness in that core task that 210 00:14:07.420 --> 00:14:11.440 again is across the board in lots of different tasks. 211 00:14:11.440 --> 00:14:15.500 Whether your checking ID's and verifying that the individual in front of you is the individual 212 00:14:15.500 --> 00:14:19.590 on the ID or your looking at x-ray images or your 213 00:14:19.590 --> 00:14:23.620 scanning a checkpoint looking for suspicious behavior. 214 00:14:23.620 --> 00:14:27.670 [Host] So I think this raises the question about talking about next steps and what else this could be used 215 00:14:27.670 --> 00:14:31.770 for, but also in terms in customization to meet these different types of needs. 216 00:14:31.770 --> 00:14:35.790 How customizable is this training? [Darren] That's 217 00:14:35.790 --> 00:14:39.860 something that we wanted to do from the beginning is be able to have 218 00:14:39.860 --> 00:14:43.960 individual organizations, whether their federal, state, local government organizations 219 00:14:43.960 --> 00:14:48.000 be able to customize their threat in the database to meet the needs 220 00:14:48.000 --> 00:14:52.020 of the stream of commerce, whatever that may be 221 00:14:52.020 --> 00:14:56.150 and the commonly emerging threats. If there's something that 222 00:14:56.150 --> 00:15:00.190 the public would see in the latest news cycle. A battle an emerging threat 223 00:15:00.190 --> 00:15:04.260 the different training organizations at the airport level can 224 00:15:04.260 --> 00:15:08.270 load those emerging threats right into the threat image 225 00:15:08.270 --> 00:15:12.310 database and start training their screeners 226 00:15:12.310 --> 00:15:16.390 that day in real time as the threats are emerging. [Host] That's amazing. 227 00:15:16.390 --> 00:15:20.410 We have another question from facebook. In learning from this technology 228 00:15:20.410 --> 00:15:24.450 is TSA loosing anything in the way of human instruction, how do 229 00:15:24.450 --> 00:15:28.530 managers balance their own instruction with the tech is telling its trainees? 230 00:15:28.530 --> 00:15:32.550 Well the on the job training is a huge 231 00:15:32.550 --> 00:15:36.600 part of the way TSA are trained. This is 232 00:15:36.600 --> 00:15:40.690 exclusive to the classroom and to the 233 00:15:40.690 --> 00:15:40.780 current monthly training and 234 00:15:40.780 --> 00:15:44.810 current monthly training and part of the initial training at 235 00:15:44.810 --> 00:15:48.870 the new TSA training academy, but what we are looking to 236 00:15:48.870 --> 00:15:52.970 do is definitely is maintain that approach that TSA 237 00:15:52.970 --> 00:15:57.000 has where they do classroom training and then they do 238 00:15:57.000 --> 00:16:01.070 extensive on the job training where the TSA is 239 00:16:01.070 --> 00:16:05.180 shadowed by a monitor and getting those 240 00:16:05.180 --> 00:16:09.200 lessons learned, that expertise from the trainee 241 00:16:09.200 --> 00:16:13.270 and from the trainer that their paired with. Another thing 242 00:16:13.270 --> 00:16:17.270 that we incorporate into the training was extensive 243 00:16:17.270 --> 00:16:21.320 knowledge and lessons learned from the very best of the 244 00:16:21.320 --> 00:16:25.400 TSA workforce. We spent some time out in the field doing cognitive walk throughs 245 00:16:25.400 --> 00:16:29.410 and structured user interviews with the top 1 to 3% 246 00:16:29.410 --> 00:16:33.460 of screeners that hardly miss a threat in training and 247 00:16:33.460 --> 00:16:37.470 then we turned that into a special training module for ID's and ID componets. 248 00:16:37.470 --> 00:16:41.490 So really trying to leverage the expertise of 249 00:16:41.490 --> 00:16:45.530 our training workflows. [ Host] So this incorporates a lot of then the user input in terms of 250 00:16:45.530 --> 00:16:49.620 what should people be looking for? Talking about the top 1 or 3% 251 00:16:49.620 --> 00:16:53.630 of screeners. Getting their input about, this is what I know, 252 00:16:53.630 --> 00:16:57.680 this is how I approach this. Really getting a human input 253 00:16:57.680 --> 00:17:01.770 into this training so it's not purely just a technology focus to 254 00:17:01.770 --> 00:17:05.780 the facebook users point about melding a 255 00:17:05.780 --> 00:17:09.830 human-based instruction, but also a technology 256 00:17:09.830 --> 00:17:13.930 technical capability. [Darren] Absolutely! This is where it is just about giving 257 00:17:13.930 --> 00:17:17.960 taking advantage of emerging technology, taking advantage of the latest in training research, 258 00:17:17.960 --> 00:17:22.020 visual search research done by the larger community 259 00:17:22.020 --> 00:17:26.030 and giving the trainers more tools in their toolbox. 260 00:17:26.030 --> 00:17:30.070 Everybody needs a minimum performance level of course 261 00:17:30.070 --> 00:17:34.130 but some people are predisposed to being good 262 00:17:34.130 --> 00:17:38.150 at that screening test and by distributing a capability 263 00:17:38.150 --> 00:17:42.180 to the workforce to give them these extra tools can increase 264 00:17:42.180 --> 00:17:46.260 national security by bringing everybody's individual performance level up. 265 00:17:46.260 --> 00:17:50.370 [Host] Awesome. So how does this apply to looking at ID's? 266 00:17:50.370 --> 00:17:54.410 [Darren] So this, the ID or identify 267 00:17:54.410 --> 00:17:58.490 portion of the system that capability 268 00:17:58.490 --> 00:18:02.500 specifically looks at how 269 00:18:02.500 --> 00:18:06.550 in this case, CBP officers were looking at 270 00:18:06.550 --> 00:18:10.640 individuals and trying to go through their progressions 271 00:18:10.640 --> 00:18:14.660 that their taught. You know, eyes to ears, nose to chin 272 00:18:14.660 --> 00:18:18.710 those sorts of things to verify that the individual 273 00:18:18.710 --> 00:18:22.790 in front of you is the individual in the ID. 274 00:18:22.790 --> 00:18:26.810 So the correction itself could be real, 275 00:18:26.810 --> 00:18:30.870 could be real but altered, could be fake or 276 00:18:30.870 --> 00:18:34.970 could be real but not the person in front of you. 277 00:18:34.970 --> 00:18:38.980 What we did in a training effectiveness evaluation for the identified portion 278 00:18:38.980 --> 00:18:43.040 was we worked with CBP to 279 00:18:43.040 --> 00:18:47.150 pilot the system and it resulted in 280 00:18:47.150 --> 00:18:51.180 those officers being able to, not only in the class time that 281 00:18:51.180 --> 00:18:55.250 they have being able to actual look at more image pairs, 282 00:18:55.250 --> 00:18:59.360 because they are able to make decisions faster with more confidence. [Host] Oh okay. 283 00:18:59.360 --> 00:19:03.400 [Darren] But they are also able to detect more impostors. So it actually was 284 00:19:03.400 --> 00:19:07.470 an efficiency increase in that they are getting now in the same amount of time being able to 285 00:19:07.470 --> 00:19:11.480 look at more image pairs and get more training in 286 00:19:11.480 --> 00:19:15.520 the same amount of time, but also actually improve their 287 00:19:15.520 --> 00:19:19.600 performance in impostor detections. [Host] So really, if 288 00:19:19.600 --> 00:19:23.610 we're increasing effectiveness and efficiency. Not 289 00:19:23.610 --> 00:19:27.660 sacrificing one for the other. [Darren] Absolutely. [Host] We have another question from facebook. 290 00:19:27.660 --> 00:19:31.740 Do TSA's have to wear something special for eye tracking? 291 00:19:31.740 --> 00:19:35.750 [Darren] No, the eye tracking technology is 292 00:19:35.750 --> 00:19:39.810 totally unobtrusive. It takes just a few seconds to 293 00:19:39.810 --> 00:19:43.900 actually calibrate and set-up. They look like 294 00:19:43.900 --> 00:19:47.930 a little ball that hooks to the bottom of the display 295 00:19:47.930 --> 00:19:51.980 and so that with a short calibration limit. Again it's meant 296 00:19:51.980 --> 00:19:56.080 to be used with or without an instructor present. 297 00:19:56.080 --> 00:20:00.110 They calibrate the system to start tracking 298 00:20:00.110 --> 00:20:04.170 their eyes and their often running into their training [unintelligible]. [Host] She almost 299 00:20:04.170 --> 00:20:08.270 looks like a heat map of where somebody looked. 300 00:20:08.270 --> 00:20:12.310 [Darren] Yeah, the hardware itself is a eye tracking, its a small device that 301 00:20:12.310 --> 00:20:16.380 is attached to the bottom of the display. So nothing that you actually wear. 302 00:20:16.380 --> 00:20:20.490 It's tracking your eyes from the bottom of the display. 303 00:20:20.490 --> 00:20:24.540 It can be in a single or double configuration 304 00:20:24.540 --> 00:20:28.540 a lot of federal buildings use a single screen configuration. 305 00:20:28.540 --> 00:20:32.560 Our current set-up at TSA checkpoints is with dual screen 306 00:20:32.560 --> 00:20:36.590 configurations so it is offered in both configurations. 307 00:20:36.590 --> 00:20:40.670 [Host] Another question, can organizations other than TSA or CBP 308 00:20:40.670 --> 00:20:44.680 use this software? [Darren] Absolutely! The system then is 309 00:20:44.680 --> 00:20:48.730 available to all federal, state and local governments. We are also pursuing 310 00:20:48.730 --> 00:20:52.730 a release to other 311 00:20:52.730 --> 00:20:56.740 private organizations, cruise lines for instance. 312 00:20:56.740 --> 00:21:00.790 And of course collaborating with our other 313 00:21:00.790 --> 00:21:04.870 governments that's we have bilateral agreements with. 314 00:21:04.870 --> 00:21:08.890 So it is available to all state, local, county and other 315 00:21:08.890 --> 00:21:12.950 federal organizations as well. [Host] So how do they get access to the training is it on the web? 316 00:21:12.950 --> 00:21:16.960 Do they email somebody? How would you recommend if someone is interested in this they go about 317 00:21:16.960 --> 00:21:20.980 clearing it? [Darren] First I'd respond 318 00:21:20.980 --> 00:21:25.030 by going to HQ.dhs.gov, I think is 319 00:21:25.030 --> 00:21:29.130 gonna come up on the page there. You can reach out to us 320 00:21:29.130 --> 00:21:33.150 we will follow up with you on inquiries about using 321 00:21:33.150 --> 00:21:37.220 the system for your specific application and 322 00:21:37.220 --> 00:21:41.320 set-up a demo of each of the capabilities if anyone is interested. 323 00:21:41.320 --> 00:21:45.350 [Host] Good and more information is about this tool is available on 324 00:21:45.350 --> 00:21:49.420 S&T's website. If you go onto our website we have a blog, 325 00:21:49.420 --> 00:21:53.530 what we call "snap shot". There are several written pieces you can use to learn about this program 326 00:21:53.530 --> 00:21:58.530 about this project. Kinda diving into a couple more 327 00:22:01.610 --> 00:22:05.700 questions. I know we have talked a little about where this technology could go and one of the exciting things about 328 00:22:05.700 --> 00:22:09.740 being in S&T is you are never quite sure where something is going to end up in the future, 329 00:22:09.740 --> 00:22:13.820 but if you had to kind of project after different iterations or 330 00:22:13.820 --> 00:22:17.830 going down some other uses. What would you like to see this technology become? 331 00:22:17.830 --> 00:22:21.880 You know 5-10 years from now? [Darren] I think as 332 00:22:21.880 --> 00:22:25.970 we move forward we want to offer additional capabilities, 333 00:22:25.970 --> 00:22:29.980 like I was saying for suspicious behavior detection and 334 00:22:29.980 --> 00:22:34.040 pre-deciding indicators other tasks that are at their core visual search tasks. 335 00:22:34.040 --> 00:22:38.140 That could leverage the eye tracker and the adaptive training. 336 00:22:38.140 --> 00:22:42.140 But even the adaptive training capabilities itself, the other core modules that we 337 00:22:42.140 --> 00:22:46.140 developed. The different methods of trainings. Exposure versus discrimination training. 338 00:22:46.140 --> 00:22:50.250 And the customizable database of threats 339 00:22:50.250 --> 00:22:54.280 could be put into a web enabling implementation. So we 340 00:22:54.280 --> 00:22:58.340 could have even those organizations that don't have dedicated training rooms to have 341 00:22:58.340 --> 00:23:02.450 a single or double screen set-up 342 00:23:02.450 --> 00:23:06.450 that they will be able to access it over the web and 343 00:23:06.450 --> 00:23:10.520 have the distributing capabilities to a workforce. [Host] That would be great. So really it's 344 00:23:10.520 --> 00:23:14.630 widely available, making it a very effective, you know 345 00:23:14.630 --> 00:23:18.660 cost effective training option. [Darren] Yes, absolutely. 346 00:23:18.660 --> 00:23:22.750 [Host] Great, unless there are anymore questions is there anything else that you want people to know about 347 00:23:22.750 --> 00:23:26.750 screen adapt and identify? [Darren] I think that its a new 348 00:23:26.750 --> 00:23:30.800 innovative capability. I think that it's definitely worth checking out if you are 349 00:23:30.800 --> 00:23:34.880 in that homeland security enterprise federal organization 350 00:23:34.880 --> 00:23:38.900 security organization, local, state 351 00:23:38.900 --> 00:23:42.950 law enforcement. Any organization thats doing ID 352 00:23:42.950 --> 00:23:47.030 verification that is looking for impostors and especially doing check 353 00:23:47.030 --> 00:23:51.050 point screening is definitely something that could 354 00:23:51.050 --> 00:23:55.110 impact the operational security of 355 00:23:55.110 --> 00:23:59.200 your particular scenario. [Host] Great. So again please reach out if you have any questions 356 00:23:59.200 --> 00:24:03.220 we're going to leave this thread open, so if you do have questions we want to answer by text 357 00:24:03.220 --> 00:24:07.280 in the days to come. Thank you for joining us for our first Tech Talk and be sure to stay tuned 358 00:24:07.280 --> 00:24:11.370 for us on facebook and twitter to find out what our next subject is going to be. 359 00:24:11.370 --> 00:24:14.520 Thanks a lot Darren. [Darren] No problem. [Host] Alright, thank you.