Placeholder Image

字幕列表 影片播放

  • sound Check.

  • 123 How's my audio, everybody?

  • Good morning.

  • Happy Saturday.

  • It's like like the old days in Saturday morning when everyone would wake up and gather around the television tube, not the YouTube and watch cartoons.

  • And now you're all here gathered around the YouTube and watching the coating train.

  • It's very weird to be here on a Saturday, especially in the morning.

  • I don't think I've ever strike.

  • Probably streamed on a Saturday before.

  • I don't know that I've streamed on a Saturday morning before.

  • This isn't a particularly great time.

  • There's something about this time.

  • It's actually quite nice.

  • I feel sort of relaxed in a bit at ease.

  • Um, and I'm very glad to be here.

  • But also, you know, the weekend.

  • It's the weekend and there's lots of family stuff in other things.

  • Going on spoke.

  • I don't know that I'll be making a habit of this, Um, but but here we are nonetheless welcome.

  • It's so nice to see so many people here.

  • I don't actually see you, but I feel your presence feel more calm today for some strange reason.

  • Maybe because I've read the marathon last weekend, Um and, um I have a very particular topic that I want to focus on today, just quickly before I go.

  • That topic I want to just thank the sponsors today for Saturday is coding Train Live Stream, which are Dash Lane and Lynn owed.

  • So if you go to dash lane dot com slash coating train or leno dot com slash coding train, you can find out about both of these.

  • Web service is Dash Lane for as a password manager and Lynn owed as a cloud server, and I'll come back in the middle of live stream and talk to you a bit more about the sponsors and show you some clips and things features that they have and tell you about the coupon codes.

  • Okay, so huh um, console log.

  • Hello world says Cybertron.

  • No cyber crime.

  • It is, but still a bit low, and Nathan in the chat is, Well, it's Ah, Nathan is a very loyal viewer who has really been getting at me to do some, uh, upgrading of the audio equipment that I have is actually quite good.

  • I believe I have a lab might hear actually a shotgun mike over there that I don't use But what I really need to do is some fine tuning of the audio.

  • I use a piece of software called Open Broadcast Studio, Actually.

  • Check.

  • I don't believe I'm recording anything to disk right now.

  • I'm gonna add that it's right here.

  • I'm gonna start recording on and the focus is also done.

  • A point.

  • You know, I I'm out of my mouth of practice here.

  • It's Saturday morning, and I just like, what's going on?

  • I should be at home having brunch and putting my reading a nice magazine or book and listening to some jazz music on the radio.

  • Uh, you use this?

  • Not a sponsor, but favorite one of my favorite products called the Stream deck, which is by tool for focusing, because I put it over there and then I walk over to the camera and I focus on it.

  • Uh, there we go.

  • I think that's probably my eyesight is actually quite bad.

  • No, I can't even see.

  • Not enough.

  • I think that's better.

  • It's the one thing I want to mention today.

  • If you have been paying attention to the coating train YouTube General, you would be aware of this.

  • But if you're not.

  • If you go to the coding train dot com slash tm for teachable machine, we'll take you to a Web page, which has three videos that I actually two videos that I released this week and one that's still sitting on the channel as unlisted.

  • Other people are asking for the audio to be a little bit louder.

  • I'm always afraid to turn it up because I'm just worried about it peeking.

  • But everybody is telling me that it could be louder.

  • So let's, um, let me go.

  • I have a little dial.

  • Here's this little dial.

  • I see it here.

  • It's called, says Gain on it.

  • It's on this audio interface.

  • It's blinking green, and if I just turn it a little bit to the right, I have now made myself a little bit louder, I think.

  • And I see it's in the yellow, but not in the red.

  • The problem is, I get very excited.

  • Sometimes.

  • Run lime street.

  • I'm going into the red.

  • But a teachable machine is a ah project from Google Creative Lab that I referred to in a lot of my Ml five GS videos, and but they released a new version of it.

  • If I go to teachable Machine with google dot com, I will be on this web page, and it is a fast and easy way to create machine learning models for your sights, APS, amore.

  • No expertise is required, and you can see that.

  • So I'm not gonna go through this right now because I have three videos that go through it in quite a bit of detail, but I encourage you to check out the video about it.

  • They actually have their own tutorials that you could watch that are much more succinct.

  • It is like insane is mine.

  • So I don't know what your preferences, but if you wanna learn how it works, I would definitely recommend you watch theirs instead of mine.

  • If you want to watch somebody embarrass themselves for 15 or 20 minutes, then you can watch mine, Um, some wonderful projects.

  • And what?

  • What?

  • I'm excited.

  • One things I'm very excited about is how they've integrated the teachable machine machine learning models with all these other libraries and frameworks, in particular the P five Jess Library with the Ml five library.

  • So I don't wanna go through this now.

  • It really relates to what I do want to do in this session, which is, uh, look at a particular feature of the ML five library on that also deals with training a model.

  • But I would encourage you to check out my videos.

  • And in particular, I would really love to see what people make with this.

  • I mean, I love that for everything, but I'm particularly curious as to whether this methodology of working this interface in the browser that allows you to train a model and then download that model or export upload that model tow work with, say, P five J s and M l five.

  • If that's something that inspires people to make creative work.

  • And so, for all any of these videos, if you click on this to the tutorial button on dhe, there's actually the one of them has you'll see there's this area called community contributions and I want to see your community contributions here.

  • So So, please, uh, because I would like to come back and look at some of them and play with them on a future video.

  • There was one.

  • So I think if I go to the maybe the snake game, yes, So, um, Mika Crew Shell, who has been, ah, wonderful supporter of the channel in contributions on Get Hub.

  • Thank you so much for adding a lot of stuff to the website and things that I really appreciate that this is the only community contribution that I know of.

  • So let's try it.

  • And what?

  • One of the things that's really interesting about the system is if you train the model to say, using your webcam, it's going to learn the context of what your webcam is looking at.

  • So let's say you train it to recognize a banana versus an orange.

  • Would that still work when I look at it here in this context with the green screen behind me?

  • Um, all right, so I'm here.

  • I am loading.

  • I think the idea here is that I'm supposed to play rock paper scissors with the computer.

  • Yeah, I guess the computer is just winning.

  • Oh, no, I want I guess it's actually detecting the knighted scissors or having to paper.

  • It detected my paper.

  • That's awesome.

  • Of course I lost paper again.

  • I win.

  • Let's see, it doesn't detect rock, but here we go thinks that scissors thinks that's paper.

  • So a couple things I would say here is that interaction, design wise, just has a little critique here.

  • I'm kind of confused.

  • So what I think I understand is that's what the computer's picking.

  • And that's what I'm picking.

  • But what I'm what I'm really confused here is about the timing.

  • So I feel there's this, like, ready.

  • But I feel like what might be helpful is some sort of like Countdown timer get into position, because when you play it, one way to play the game is rock, paper, Scissors says, Shoot like that kind of thing.

  • So I wonder if there's some kind of way that it could step me through the process a bit more.

  • And maybe the visual language of this being bigger is meaningful and the like.

  • The Emojis is like, What's the computer versus me?

  • Are also, I think, quite effective.

  • But something about the difference in sizes and the placement.

  • I like that the video is there but kind of faded into the background.

  • That's kind of I think, a nice idea, but I think that maybe there could be some other ways to think about how to do the interface I also am really lost in finding myself because I think this is cropped and I would I would always I would mirror the video because it just makes interacting with things much easier because oh, no, it is mirrored.

  • It is mirrored.

  • Stephanie mirrored.

  • I just was Oh, I see.

  • Maybe it's doing that on purpose so that if I put so that there we go I see this is what it wants me to do.

  • So maybe there could be there could be a bit of a stage where it lets you kind of get set.

  • And it seems there's a moment also where it's like freezing.

  • And that's where I find that very distracting eso.

  • I kind of want to see it always great.

  • So I want to see more of these.

  • Oh, Mika's in the chat.

  • The timer is the pink bar at the top.

  • Okay, I got it.

  • I see.

  • I see.

  • And Simon is saying that there have been a bunch of community contributions from the interactive drawing video.

  • So maybe I will.

  • Let's go, Let's go.

  • Let's go take a quick peek at that.

  • One thing I need to do so I have this iPad here where I can play and the cold.

  • Let's conjecture.

  • Okay.

  • Okay, everybody, settle down.

  • Settled out.

  • Um, it only has 2% battery left, so I'm going to unplug Go plug over here.

  • So I won't have any music for a little while when we let this charge up, but it'll charge up by hopefully the time that I need music.

  • All right, let's check out what else is happening on the channel here.

  • There's the interactive drawing was one of a recent video that I can take a look at and also the coal lots conjecture.

  • I haven't had the time to do another one of these cabana videos, which I'm very sad about.

  • Ironically, I probably would have done one this morning if I wasn't coming here.

  • Do this live stream, but too many things to do.

  • Okay?

  • I don't see any community contributions for this one yet.

  • I'm shocked.

  • I'm shocked.

  • I'm shocked.

  • After committing contribution for this video, let's check out the cool s conjecture.

  • That definitely have been a much whoa.

  • There's a lot.

  • Amazing.

  • Let me just do a quick click through.

  • I'm just gonna open them all up.

  • This one for some reason, doesn't have a link.

  • Let's zip in that.

  • This is probably a bad idea what I'm doing.

  • Let's let's just let's just look through these, okay?

  • Of course.

  • Little Rainbow Cola's conjecture.

  • Visualization.

  • Gotta love that.

  • This is by a sequential chaos.

  • Now we've got one.

  • That is by Al Barco.

  • Looks pretty similar to mine, but there's probably something different here that I'm not aware of.

  • Let's see if it says growing visual ization.

  • Oh, no, no, no.

  • This is this is not similar than mine.

  • I made a version of this like this myself just to make a gift to post on Twitter, but this one is actually animating it.

  • I forgot that my video drill doesn't animate it, so that's wonderful.

  • Thank you for that.

  • Um, cola tree refresh.

  • Oh, sod.

  • Where's the tree?

  • Why is it showing me all this nonsense?

  • Okay, I'm not sure what's happening with this one.

  • We can click on looking at the code and try running it here.

  • You have to click.

  • I probably to click or something.

  • No, I'm not sure what's going on here.

  • Um, Donna, we let me know.

  • I'm happy to try showing it again.

  • Um, this is Ah, Cole.

  • Lots visualization from a spectral flame Added code for animated and static drawing.

  • So one thing I would certainly recommend if you can, let's see, is thio to read me file with, like, a gift animation or a screenshot that scroll down the page says Simon, Let me go back to that one.

  • The coal, it's growing tree.

  • Let's go back to Ah, it was just It was just taking up a lot of space.

  • There it is, Another growing one.

  • And I like the color selection here.

  • The green.

  • This is a good team.

  • Trees.

  • You have a major team trees donation.

  • Please go ahead and do that.

  • Um, who hears it?

  • Really?

  • Net.

  • Look at this one.

  • I like this from super Rambo.

  • Um, this is Kolat.

  • Sunray, Sunrise from Anna.

  • Rog has rocked.

  • This is quite lovely.

  • This was I saw this on Twitter.

  • Onda.

  • Lot of people had sort of fun names for this kind of supernova like design.

  • And I like this use of dat gooey here.

  • Looks like I could probably, um, try changing various parameters and I don't know if Oh, and then if I click, I guess it will.

  • It'll it'll Every time I click, it will reset and rerun.

  • With these new parameters, I really want to do some or prince printing of these and makes them so that's our plotter stuff.

  • Gonna come back to that today.

  • This is lovely.

  • Um, okay.

  • Um Kolat, Swede, this is interesting again.

  • Another processing one.

  • There's this one.

  • I really want to kind of run this.

  • So I've drive like, I guess, I guess to run one that's in processing, I have to just download it.

  • And now if I'm going to do this, I should also go back to spectral flames, because that's only fair.

  • Um, a lot of stuff in this downloads director, but since there's a fish, I got to know what this fish is.

  • Oh, look at that.

  • I love this.

  • So I love this.

  • I What?

  • I what I love about this, which surprises me in a way, is using I suppose this cold lots conjecture, visualization in scene.

  • So you could imagine this being a to D game that you're playing with your moving of drug money.

  • You're the fish and you're controlling the fish in the fish is to eat or swim or whatever, but all the while little is, you know, the trees at the seaweed at the bottom was all coming from the coal.

  • It's conjecture.

  • Um, okay, let's, uh Kolat click.

  • See tree?

  • I guess so.

  • This looks like maybe it just picks a different angle in a color scheme each time, which was really, really nice.

  • So nice to see these on these variations.

  • Um, another rainbow one.

  • Thank you, Louise.

  • Still l and we're seeing this also on code pen.

  • So it's nice to see people try different code editors and frameworks.

  • Um, Eric Rovell It posted something about trying a 60 degree angle.

  • And it looks like the image I'm seeing over here on my slack chat screen is showing a very like, hexagonal pattern, which is nice.

  • I don't know if that's going to show up another one here.

  • Um gear Elice dash by diplo Focus.

  • Oh, wait.

  • I think I made everything tiny.

  • There we go.

  • Oh, I like this.

  • I like I like this idea of these sort of, like, dash lines here.

  • Um spectral flame.

  • We're gonna try that.

  • Let's try spectral flame.

  • Let's bring this here.

  • I hope it's getting this here and try this.

  • No, that's the other one.

  • Not this one.

  • There we go.

  • Who?

  • I like this.

  • I like the layering of this.

  • What's different?

  • I guess it's doing huge sections at once and then varying the color and kind of blending it.

  • It's quite o This is quite enjoyable.

  • Needs a beauty needs this needs full screen in music.

  • So let you enjoy this for a minute.

  • Oh, I still have my Michael.

  • And I thought I knew it in my mind.

  • I was thinking a break there.

  • Give myself a break.

  • That was beautiful.

  • I'm, like, probably some tea over there.

  • Okay.

  • Back.

  • Hello?

  • This is lovely.

  • Ah, beautiful.

  • Just beautiful.

  • Um, okay.

  • Refresh the repo.

  • Look at this.

  • Here's a nice read.

  • Me.

  • I'm also listen.

  • Okay.

  • The version of number file.

  • Using the angle of pie, divide by 13 to the right and pi divided by 20 to the left.

  • They have these long lines with bumps on the underside of them.

  • Okay, I think if I'm right spectral flame, maybe you also commented in the actual YouTube video.

  • Somebody wrote a really excellent commentary about somewhere that I read.

  • It was on Twitter on the YouTube video.

  • Or maybe it was uh oh, baby.

  • This was the conversation we had on Get Hub.

  • But on maybe with Spectra flaming apologies, it was somebody else who really did it turn a deep dive and figuring out what were the properties of the visualization from the number file video and how those were different from the ones that I the version that I created on looks like there's a really nice explanation of that here and played around with some more and got this colorful artwork and equals.

  • It's always hard to read.

  • Five million Change the color plow to include all colors.

  • Wow, look at that.

  • That is quite something technique Collins is asking, um, where can I get the documentation for the quick draw project?

  • So I'm not sure exactly what you're referring to.

  • But if you're referring to the actual quick draw data set that is also from Google Creative Lab, and there's a get hub repo that has all of the documentation of the quick draw data set.

  • If you're looking for information related to my video to twirl, that makes use of a machine learning model called Sketch are in end, which was trained on the quick draw data set.

  • Then you can find that at the coding train dot com.

  • If I go to this the Challenge page and typically I mean, sometimes things are missing here, and if they are, you can file an issue or actually submit the link.

  • But all of the relevant links related to the material that I use should show up here other videos that are related to show up here and then community contributions.

  • But if you're looking for the code, you could find that here and a lot of time.

  • My goal is to always have JavaScript code processing code and a link to the version running in the Web editor.

  • But this one has no processing code because I don't currently have a Java.

  • Uh, there's not an easy way to bring the sketch, aren't and model into Java.

  • It's definitely possible, but it's not saying that I've spent some time doing um, I added this overlay thing.

  • I haven't been any new members or anything joining yet.

  • I'm just curious if something's gonna pop up this that will.

  • Cody train will ask DT six.

  • Will you try que learning in the near future, so I would answer yes to that question.

  • If you could take the word near out, it's really it's I'm having a tough time.

  • I'm definitely doing a lot on making content.

  • And I made this list and I definitely don't have it.

  • Simon probably hasn't hit somewhere at the beginning of the semester of kind of the things that I wanted to do.

  • This semester, I've come to very few of them.

  • I really got into some of them, but under very few of them, acu learning is on my list of things that I would like to tackle.

  • It's not something that I have a lot of experience with.

  • Ah, researcher here at Y you named Aidan Nelson has made a bunch of Q learning examples with P five.

  • I can try to find those and link those David Seiders.

  • Well, that looked like that overlay works.

  • I would have to refund your money.

  • Thank you.

  • Oh, that's sad.

  • Do what I need to do to get that to work.

  • Uh, well, um, there we go.

  • Thanks.

  • Simon is now sharing with me on my list.

  • Only if there's an easy way for me to pull that up.

  • I'm on screen.

  • Um, but and Dan man asks, are you going to make videos about Are making our personal teachable machine framework.

  • Ah.

  • Okay.

  • Well, let me cover this.

  • And then some people are asking about whether the teacher will machine.

  • I'm sorry.

  • This is not a call to ask people to give super chats.

  • I apologize.

  • I didn't know it was not really my intention.

  • Thank you.

  • Very generous of you.

  • And it's very much appreciated.

  • Um, and I thought I had a little overlay things about people names up for their messages, but apparently that didn't work.

  • Um, so I forgot about the questions.

  • Let me live so many good questions.

  • I'm trying to trying to get trying to answer these.

  • Okay, I'm gonna I'm gonna actually scroll the chat, which will stop it automatically scrolling.

  • Okay.

  • Are you going to make videos about making our personal teachable machine framework?

  • So?

  • So I actually have a MME.

  • Well, look at all.

  • What?

  • It interesting that the pops up this weird.

  • Um, what I'm looking for is the, uh, um, YouTube playlist.

  • But it's kind of interesting.

  • Um, I guess from the amount.

  • This This is what this is the link.

  • I'm looking for us.

  • Apologies.

  • Oh, um, I want the playlist, though.

  • Uh oh.

  • The internet is a rough place.

  • Um, I don't have this playlist linked in an obvious place.

  • Let's try, uh, with the chances on, like one of these probably not website working on the website.

  • No.

  • No.

  • Ah.

  • Issue to get hubs like and pull a Thank you.

  • That's an excellent way of doing this.

  • I have a lot of content, but I guess I have to go.

  • Do I have to do what I didn't want to do, which is actually just go to my channel.

  • How are there 404 7 people watching us right now?

  • Um, and look for, um, playlists.

  • Oh, my God.

  • You know, it's bad when I can't even look at this private like videos.

  • I want that to show up.

  • I'm not, am I need everything else.

  • I'm logged into my like, main account.

  • That's not my intention.

  • Uh oh, God, this is this is not going well.

  • Um, created playlists I'm looking for is it's gonna happen for me.

  • Yeah, but who?

  • Ernest.

  • Ernest?

  • I found it, huh?

  • That's very sad.

  • The question was, Are you going to make videos about making our own personal teacher?

  • Lucien Framework?

  • If you want to do that, you can do that with the the discussion in my ML five video about transfer learning with the feature extractor and then running your own feature X factor classification example.

  • This is basically a bit more that's that in terms of the tactical material that you would need to actually create the training interface and control the training process yourself.

  • Maur closely.

  • I'm so sorry that took so long to pull that up.

  • And what I want to do today is actually add make content to go through a bunch of examples in this live session that will get edited it down into videos that will end up in this playlist about training a neural network model.

  • Um, all right, so the other questions were the teachers machine exports are currently broken.

  • Is that being looked into?

  • I don't know.

  • I do know that there is a a little bit of a source of confusion.

  • A bunch of people posted that they had an air like a fetch error, and maybe I can just touch on that really briefly.

  • So if I were, let's just pretend for a second that I just trained a model and I went here to export model.

  • There's no model train.

  • So you're going to get Once you've gone through this whole process, you're going to get a U R L that's here.

  • And then you're going to see you here.

  • You're also going to see a code snippet.

  • What's confusing is the girl that teacher machine gives you does not include.

  • It's just the path to the all the model file.

  • Excuse me, but it does not include model dot Jason, which ml five expects.

  • So if you were copying and pasting, yeah, thanks.

  • Thanks.

  • I can tell if you're copying and pasting this code the Ural that you have for the model that gets stored in variable and then you have to upend modeled up Jason for it.

  • I was seeing people have that error.

  • Um, there's a YouTube problem with emotes on streams.

  • Spamming like 10 and Moz could get your Google count suspended.

  • All right, so those were some questions that I wanted to answer.

  • I think I better get started with some content.

  • It's already 11 o'clock on.

  • I'm going to do my sponsor segment.

  • And around 10.

  • 30 11.

  • 30 travel back in time.

  • Eso someone keep me on track here.

  • And I think this is gonna go until 12.

  • Or 12.

  • 30.

  • I mean, at this point, most likely 12.

  • 30.

  • Okay.

  • How's that?

  • Let me see how my Oh, and, um, let's go to the, um let's since I asked for it.

  • Let's pull up.

  • Um what?

  • Get hub dot com slash website.

  • What is wrong with me?

  • Coding train slash website issues and to do list.

  • Okay.

  • Um wow.

  • So this is Let's let's just take it.

  • Let's do a little accounting here.

  • RTP done.

  • More rows.

  • Done.

  • Colette conjecture Done.

  • Um, ticked.

  • Not note.

  • Nope, None of these.

  • None of these done.

  • We could add, by the way, team trees edited version.

  • I don't know if it makes sense to edit that into something.

  • That's sort of a discussion with Mathieu.

  • Uh, so this is what I'm gonna do today.

  • Excellent.

  • I sort of did this.

  • I sort of did this, but I changed Drew.

  • Making the video tutorial.

  • We discovered a lot of problems with library and change going to redo this tutorial.

  • I did this.

  • I haven't done any of these.

  • I'm done.

  • A These October fest is over.

  • Streaming improvements, huh?

  • Night, Baden says Nike.

  • But that should say night, Bott, I believe discord on, then 2020.

  • So this stuff in 2020?

  • Okay.

  • All right.

  • So this is good.

  • I'm not doing terribly late in space stuff.

  • I'm really excited to do with Runway ml.

  • So have some plans for that.

  • This I definitely want to dio This is a collaboration with a YouTube channel called Practical Engineering.

  • I need to get back to practical engineering about that.

  • And today's topic will be this.

  • I don't think they'll be a coding challenge today.

  • I'm looking at this.

  • Um um, this would be fun to d'oh.

  • Um, but today I think today this is today.

  • Mm.

  • Five neural network.

  • Um, okay, so let me Let's see how my charge is going with this 6%.

  • Boy, that really didn't do very good.

  • I guess I could try to get it, keep it plugged in, have anything that this could connect Thio.

  • Not really.

  • There's no plug closing up here.

  • Just looking at my notes for my class somewhere.

  • Okay, I think I'm gonna do this stuff.

  • So in case you're wondering what I'm looking at, I'm teaching a course at N Y.

  • U this semester called Introduction to Machine Learning for the Arts.

  • And I've been preparing a lot of material and doing a lot of teaching in the course about it.

  • And I haven't been able to keep my my plan host to be making videos for the course all semester long and and to some extent I've been doing that.

  • But I got way behind, so I'm kind of going back to remind myself what I did in class and sort of deciding how to how that is going to feed into what I want to show in this particular set of videos.

  • Um, and so I think I think I'm ready to go.

  • And the other thing?

  • I want to do what I've up one upgrade I've made to my recording.

  • A system.

  • Uh huh.

  • Um, is that have, um, way.

  • And I'm about to turn this on of recording all the different feeds to disk separately, green screen, the laptop screen and the white board.

  • And so this If I'm gonna edit If my say I If Macha, who does the video editor for the coating train is going to edit all this stuff together, it's actually really helpful to have all these of separate things in case we want to, like, add some more content and fix some things up.

  • Okay, so I'm just looking in the set, so I should So I need to record.

  • Um, output Shoot.

  • Sorry.

  • Give me a second to your apologies to everyone that I'm doing this during the stream, but, uh okay.

  • Wait, No, no, no, no, no, no, no.

  • Okay, so output to is the white board.

  • Output three is the green screen.

  • Output four is the laptop.

  • And so if I go to my multi quarter and I say I want 23 and four, um, now I'm going to start recording, and I'm recording everything.

  • Okay.

  • Uh, all right.

  • So I'm trying to think of how I want to go about this is the nice things, so that this is good and bad.

  • The good news is I have Maur possibilities for creating higher quality edited versions of the lifestream later, because I now have the capability to even if I'm showing you things and talking about suffering Lifestream, I can replace the background with, um, different content or more zoomed in and highlighted content, or even, like other animations and other things.

  • So I could mention something, not even show it, and then show it later.

  • So that's the good news.

  • The bad news for you is that I don't want to fall in the trap of then just not ever showing anything in the Lifestream.

  • And I don't want also make the process of editing and putting together the videos so onerous that it becomes so slow.

  • I'm also like getting a lot of feedback from my monitor, So let me just mute this, Okay?

  • Great.

  • Um okay.

  • So, um, what I'm going to talk about today is and Ml five neural network class.

  • Hello, and welcome to another Beginner's Guide to Machine Learning with Ml five jazz video.

  • In this video, I am going to look at a piece of functionality in the Ml five library called.

  • It's kind of weird noise outside the room.

  • I also hate that this light reflects in my glasses.

  • I hear by the way, I don't have any buddies from the film industry.

  • But I hear that often actors or performers in films will get these special, non reflective, like lenses glasses because this is a common issue.

  • Obviously, if I had the lights above me or something that could also get around this issue are.

  • But is that a thing?

  • Because I'll go and buy those.

  • You could go get new.

  • I need new glasses.

  • Anyway, I need a new prescription.

  • Oh, spectral piano is off, I suppose.

  • Okay.

  • Wehrli noise.

  • I really wouldn't do that.

  • Worldly noise stuff.

  • Also, I should add to this list the nothing more arose.

  • The star Rose, maybe.

  • Okay.

  • Hello.

  • And welcome to another Beginner's guide to machine learning video tutorial with Emily five.

  • Yes.

  • Very excited about this one.

  • Typically excited about the drills like make, But this one I'm particularly excited about because I'm gonna look at something that has recently arrived in the Ml five Jazz Library.

  • So first of all, use version 0.4 point two or more recent version Perhaps, but that's the version I'll be using in this video on dime.

  • Wanna look at this functionality and ml five library called Ml five Neural Network It is a function and Emma five, that creates a empty or blank, so to speak, neural network.

  • Most everything that I've showed you in this video, Siri's so far has involved loading a pre trained model.

  • So a neural network architecture that's already been trained with some data.

  • And in this video, I want to look at making an empty a blank slate, configuring a neural network, collecting data, training the model and doing inference and the context that I want to look at that is with really time Inter activated.

  • So I'm gonna come back and maybe use a more traditional data sets.

  • There's a data set that that's on the ML five examples with the Titanic survival data set.

  • I have the data set from my color classifier, Siri's, so I'll come back and show you some examples of those as well.

  • But in this first video, I just want to do something very generic, which is create a blank neural network, use mouse clicks to train it and then then move the mouse around for it.

  • To make guesses are predictions.

  • And that might sound like a weird thing to do and hopefully start to make sense as I build the code and step through all the processes.

  • All right, so that was my beginning opening discussion.

  • I did exactly what I thought.

  • It's sort of a problem, which is that I didn't show you anything like I didn't like, click and browse around the website and find stuff.

  • But I do want to add that in later.

  • We'll see how that works.

  • All right, um, I do see that there is a suggestion about the audio context in Java script.

  • It won't be happening today.

  • I appreciate everyone's enthusiasm and wanting to have their idea shown or talked about, but I'm on a particular path, and I'm just doing the topic that I'm doing in the absolutely computational geometry is on my list.

  • Anti reflective coating supplied on their glasses.

  • I thought I had that, but I will get it.

  • I'm gonna go get new glasses.

  • It's another thing for my upgrading.

  • How bad is the reflection right now?

  • Because there is a way that I can turn the light away for me.

  • And if you walk the teachable machine, Siri's the third video Has the light turned away from me, and it doesn't reflect but then else of this way of shadow on my face, I don't know.

  • All right, um, so I'm gonna go to this page, Who?

  • So this shouldn't say come currently in development, cause it's definitely here.

  • I'm just gonna scroll it down.

  • Um, I was gonna I want to talk about Rebecca Fi Brink and Ueki Nadir.

  • So, um, let me talk about that.

  • Um and, uh, all right, the reflection doesn't bother.

  • Okay.

  • Um, all right.

  • So I also want a highlight for you the Ueki Nader project, which is a free, open source piece of software created by Rebecca Free Brake in 9 2004 Training machine learning models.

  • And I would especially encourage you to watch Rebecca Fi brings talk from the i o conference in 2018 where she talks about creativity and inclusion and machine learning, and goes through some demonstrations of with Ueki, Nader and processing and other pieces of software.

  • So a lot of the work that I'm doing with Ml five is is entirely based on recreations of many of the example demonstrations that Rebecca Fi Brink made and has done research about for years and years with the WEC An Eater project.

  • So in fact, the examples that I'm going to make in this video and the next one and the next follow ups are direct ports in a way of some of the original Ueki, Nader and processing examples.

  • But I'm gonna do it all in JavaScript in the browser with P five and the Ml five Jazz Library.

  • Um, I think what I'll say also is.

  • And, um, there's also a fairly lengthy history of creative artists using real time a training machine learning models in real time.

  • Let me see that.

  • Um uh, there's also a long There's also a fairly lengthy history of creative artists training machine learning models in real time to control musical instruments of performance.

  • Ah, visual art piece.

  • And I would encourage you to check out some of these projects Mart lit by Michele, the guy from the waters by an hedge.

  • This is not a fair amend by Guillermo Montesinos in Sofia Suazo and the eye conductor by under s rest guard that I will link to in this video's description for inspiration and ideas.

  • Um, okay, so now, though, um, let's see here.

  • Um let me just quickly go do the getting started.

  • Copy this.

  • I just want to get the library in the p five sketch.

  • Okay, here we go.

  • Um, And how am I on time?

  • 11.

  • 20.

  • All right.

  • Uh huh.

  • Say, Levy, Um, am I like, I don't think this camera is particularly level, but it's fine, right?

  • You can let me check the focus at least.

  • Um how's this whiteboard shot?

  • Can you see that house?

  • The focus.

  • I can't tell if I should if I should.

  • I'm afraid to touch this camera.

  • It's very loosely mounted to the wall.

  • I can make it a little sharper.

  • I think that's better.

  • I think I just improved it.

  • Um, okay, I shouldn't.

  • Well, never mind.

  • Okay.

  • Um, yes.

  • I'm going to definitely refer to my former explanations about neural networks in general.

  • Thank you, Peter.

  • Um, Okay, uh, our guide for figuring out how to write the code is going to be the ml five website, and there's a page on the ml five website for the neural network function.

  • But before I start diving to the code, let's take a minute to talk about what a neural network is now.

  • By no means am I gonna be comprehensive about this?

  • At this moment?

  • In fact, I'm going to give you a very zoomed out, high level overview, and I would refer you to the three blue one brown video.

  • Siri's what is a neural network which is one of the most excellent videos on the Internet To just quick to succinctly explained in great depth what a neural network is.

  • And I have gone through many different playlists.

  • I guess I should be doing this in front of the green screen kiss.

  • I want to show any of this stuff about Yeah.

  • Oh, just do it over here.

  • What, you gonna say this again?

  • Now, by no means am I going to give you a comprehensive explanation Now, by no means in this video am I going to do a comprehensive deep dive into what a neural network is and how to code one from scratch.

  • I will refer you to many other wonderful resource is where you could do that deep dive, starting with the three blue one brown video.

  • What is a neural network and some of the subsequent ones?

  • I have also have a 10 to 15 part video.

  • Siri's where I build a neural network from scratch in Java script based on a particular book called Make Your Own Neural Network, that is, in Python.

  • I have other videos were that our guides around machine learning concepts where I talk about different kinds of neural networks.

  • So I will link to all of those in this video's description.

  • But here, I'm gonna use the white port over here just to give you a very zoomed out, high level overview of what I'm talking about.

  • So a machine learning system, in the most basic sense involves inputs and outputs, A classic example of a data science approached a machine learning might be.

  • We want to have some sort of machine learning model that can predict the price of a house based on some set of factors.

  • So it might be the inputs in this case might be the number of bedrooms and the latte, latitude and longitude of where it is in the world and the number of square, the square footage or square meters the size of that house.

  • Those inputs would be fed into the machine learning model, and the output would be a number ah price on this, by the way, is called a regression, which sounds like a sort of terrifying term.

  • And somehow I have to be a PhD and statistics understand regression.

  • But what I mean in this case, a regression is the output is some continuous number.

  • Uh, it's a price.

  • It could be zero.

  • It could be 100.

  • It could be 1.5 million.

  • It's some number, which is different than the output being a classification, meaning it is one of several categories.

  • It's a cat or a dog or a turtle or its Class A or Class B or Class C.

  • So if you've watched my recent teachable machine video, Siri's about training an image classifier, a sound classifier.

  • All of those examples are classifications outputs.

  • The output that you get is a set of confidence scores and categories.

  • Um, interestingly enough, I started with this discussion, saying I was going to give you a high level overview of what a neural network is, and I haven't even begun that because I started talking about a high level overview of the machine learning pipeline.

  • The process.

  • So, in a way, that's what this video is also about, and the way that I think might be ineffective.

  • Yes.

  • And an effective way that I might be able to demonstrate this to you is Toe kind is to find the most drop dead simple, almost trivial scenario to show you all of those pieces.

  • Let me.

  • Yeah.

  • This is just the corner of the room for people who are asking why, boy, something about the whiteboard today Very hard to race.

  • I need a whiteboard cleaning system.

  • Okay, this barker is pretty bad.

  • Also, just this one's better.

  • Yeah, that's what's better.

  • Um, so I will get to talking about what a neural network is.

  • But for the moment, I'm actually gonna, like, stick with the sort of full pipeline of a machine learning project.

  • You really start this whole thing over?

  • Don't like my explanation.

  • Um, for what?

  • But whatever.

  • Let's just keep going.

  • All right?

  • I'm thinking I know exactly what I want to do, but I'm just kind of confused about this roundabout way that I'm doing this.

  • So let's say for a moment that the goal that I have is to train a machine learning model to use my body as the input.

  • So maybe how I'm moving my arms and legs and head.

  • That will be the input and the output would be a musical instrument, a note that's being played so I could somehow play different notes based on how I moved my body.

  • This is thistles, a scenario that has been This is a scenario that Rebecca Fi bring covers in great detail in her course about a machine learning on Ueki.

  • Nader Forget what it's called machine learning for artists and musicians.

  • This is This is an area that's covered in great detail in Rebecca Fi brings course machine learning for artists and creators.

  • What musicians know.

  • How can I get you know what that's called?

  • The worst, uh, artistic musicians.

  • Yeah.

  • Okay, I know you can't see it, but I'm recording it.

  • This is a scenario that's covered in great detail in Rebecca Fi brings course machine learning for artists and musicians.

  • A very simple one way that I might boil this idea down into its very simplest version is Think about a two D canvass and it's very convenient that I'm using p five because that's the thing that exists in P five J s.

  • I have a two d canvass And what I'm going to do is I'm going to say there's a mouse in that canvas and the mouse is going to move around the canvas.

  • And based on where it is, it will play a particular note.

  • Now, of course, I could do this with if statements very easily.

  • And what what's the thing that, um, the Kyle MacDonald quote?

  • Uh, Kyle MacDonald.

  • I think I actually put it in my syllabus somewhere.

  • Um, um, it's in this talk.

  • A weird intelligence, uh, is your hand if you're interested in art Nice.

  • Where some people have a heart beat something.

  • This is a really good talk, by the way.

  • It's, um, about an online request you once made about health.

  • This is great.

  • Quote that.

  • I want a reference using.

  • I knew it was generated by the I guess, production of analysis.

  • And I guess I missed it sometimes.

  • Like her Thio create sometimes.

  • Take something.

  • Fine people.

  • Are you I things like Ooh, I'm not gonna find it.

  • I'm gonna give up.

  • Um, okay.

  • Where is that from?

  • Shoot.

  • They don't know what I'm talking about.

  • I would like to reference it.

  • Let me give me give me one second toe.

  • Look for this.

  • Kyle MacDonald.

  • What is a machine learning?

  • And this is what I'm looking for in this medium article.

  • I really like this explanation.

  • I don't think I'm gonna find it.

  • I'm just gonna have to say it in my own words.

  • Instead of quoting Kyle, it was actually, um might actually have been a discussion here.

  • Um, hold on.

  • I know Kyle.

  • Oh, it's this one.

  • It's this talk.

  • Maybe or not.

  • Okay, I give up.

  • Sorry, everybody.

  • So many things around the world to look at.

  • Okay, I give up, I give up.

  • Sorry, I'm wasting way too much time. 01:02:24.170 -->

sound Check.

字幕與單字

單字即點即查 點擊單字可以查詢單字解釋

A2 初級

現場直播#188:ml5.neuralNetwork (Live Stream #188: ml5.neuralNetwork)

  • 0 0
    林宜悉 發佈於 2021 年 01 月 14 日
影片單字