WEBVTT

00:00:00.840 --> 00:00:03.199
the microbit machine learning tool lets

00:00:03.199 --> 00:00:05.960
you train test and improve a machine

00:00:05.960 --> 00:00:07.879
learning model that can recognize

00:00:07.879 --> 00:00:09.960
different actions or

00:00:09.960 --> 00:00:13.040
movements here's a quick tour of how it

00:00:13.040 --> 00:00:16.880
works you'll need two microbits you'll

00:00:16.880 --> 00:00:19.240
hold or wear the first

00:00:19.240 --> 00:00:21.439
microbit follow the instructions on

00:00:21.439 --> 00:00:25.400
screen for setting up microbit

00:00:26.080 --> 00:00:29.279
one then unplug it and attach a battery

00:00:29.279 --> 00:00:31.960
pack if you have a wrist strap you can

00:00:31.960 --> 00:00:35.480
attach microbit one to your

00:00:36.480 --> 00:00:39.360
arm the second microbit stays connected

00:00:39.360 --> 00:00:42.360
to your computer with a USB cable follow

00:00:42.360 --> 00:00:44.200
the instructions on screen for setting

00:00:44.200 --> 00:00:47.000
up microbit

00:00:47.559 --> 00:00:50.239
2 at the bottom of the screen you'll see

00:00:50.239 --> 00:00:52.840
live movement data from microbit 1's

00:00:52.840 --> 00:00:54.280
accelerometer

00:00:54.280 --> 00:00:57.120
sensor each colored line represents a

00:00:57.120 --> 00:00:59.440
different direction or Dimension you're

00:00:59.440 --> 00:01:01.399
moving the microbit

00:01:01.399 --> 00:01:04.560
in Step One add some

00:01:04.560 --> 00:01:07.320
data decide which actions you want the

00:01:07.320 --> 00:01:09.000
microbit machine learning tool to

00:01:09.000 --> 00:01:12.119
recognize clapping and waving are good

00:01:12.119 --> 00:01:13.479
ones to start

00:01:13.479 --> 00:01:17.560
with name your first action then start

00:01:17.560 --> 00:01:20.159
moving click the red button to collect

00:01:20.159 --> 00:01:23.159
your first sample of

00:01:25.079 --> 00:01:28.520
data collect at least three samples of

00:01:28.520 --> 00:01:31.840
your first action

00:01:34.399 --> 00:01:38.680
and do the same for at least one other

00:01:42.600 --> 00:01:46.320
action step two train the model this

00:01:46.320 --> 00:01:48.159
means that the machine learning tool

00:01:48.159 --> 00:01:51.040
makes a set of rules to estimate which

00:01:51.040 --> 00:01:55.360
action you're doing when moving microbit

00:01:55.640 --> 00:02:00.399
one step three test the model try making

00:02:00.399 --> 00:02:03.799
each of your actions the screen shows

00:02:03.799 --> 00:02:06.039
which action the model estimates that

00:02:06.039 --> 00:02:09.200
you're making the percentage number is

00:02:09.200 --> 00:02:13.319
how confident or sure the model

00:02:13.319 --> 00:02:16.280
is now try it yourself and see what

00:02:16.280 --> 00:02:17.959
different movements you can train the

00:02:17.959 --> 00:02:20.800
microbit machine learning tool to

00:02:20.800 --> 00:02:23.800
recognize