August 28, 2018 • 19 mins
New to Numenta? Bit of a beginner in brains? Then this episode is for you. In their 20 minute conversation, Numenta On Intelligence Podcast hosts Christy Maver and Matt Taylor break down the top 5 things you need to know about Numenta, including key resources to learn more and how you can get involved.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Christy (00:00):
Hi, this is Christy Maver.
(00:02):
And I'm Matt Taylor and you're listening to the Numenta On
Intelligence podcast.
Today's episode is designed forpeople who might be unfamiliar
with Numenta and our work.
Hopefully this episode willpique your interest and entice
you to follow along with whatwe're doing.
If you want to stay up to date,the best way to do that is to
sign up for our newsletter.
(00:25):
You'll find a link to subscribeon our website at numenta.com.
We typically send a newsletterevery one to two months.
Basically, anytime we have newsto share about upcoming events,
new papers we've published, andpartner updates, just to name a
few.
Subscribe today to make sure youdon't miss any Numenta news.
(00:45):
Thanks for joining us.
Hi Christy.
Hi Matt.
How are you?
I'm doing great.
I was thinking about the lastepisode, which was episode one
we did with Jeff Jeff Hawkins,our co-founder.
Two parts.
We're really bad at numberingepisodes, so we have episode
zero, episode one, part one, andepisode one, part two, and now
this is episode two.
This is episode two slash three.
(01:07):
Makes total sense.
Anyway, that last episode wentreally deep and we just want to
let everybody know that it's notalways going to be like that.
Right.
So Matt and I were thinkingafter listening to the last
episode, Let's do an episodethat's like the opposite, not a
huge deep dive, just a shallowskim on the surface of what it
is.
So maybe if you're new toNumenta or a beginner in brains
(01:32):
or just mildly curious, this isa great episode to start on for
you.
Especially if the last one justwent way over your head.
Don't give up on the podcast.
We're going to mix it up.
We're going to have someinteresting interviews with
neuroscientists and interns andceos coming up.
So what we want to do in thisepisode is we want to break down
the top five things you need toknow about Numenta,
Matt (01:54):
The first one being what is our mission?
Why do we exist, what are wehere for?
(02:00):
What are we here for Matt?
Well, I'll tell you what we'rehere for.
When I first heard about Numentais right after I read the book
On Intelligence by Jeff and thatreally affected the way I
thought.
So I looked it up.
I looked at what Jeff Hawkinswas doing and I saw that he had
founded this company calledNumenta and their mission was to
understand intelligence and Ithought that was just an amazing
(02:23):
mission for a company in SiliconValley to have and that is still
our mission even now over 10years later, is to understand
how intelligence works in thebrain.
Christy (02:32):
Yeah.
And you know, we often talkabout how we actually have this
dual mission of number one tounderstand how the brain works,
understand intelligence, andthat is very much a scientific
mission.
Brain theory driven, deepneuroscience details.
Matt (02:50):
That is number one by Jeff because he wants to know, like
honestly because he's curious,right?
Christy (03:00):
I think more than just Jeff wants to know.
Matt (03:01):
Yeah, I mean it's, there's a lot of passion behind our
mission because another thing Ilike about Numenta.
Christy (03:06):
So the first part of the mission is a scientific one.
There is a secondary mission,which is to apply those
principles of intelligence tosoftware to enable intelligent
machines.
So there is a machineintelligence, AI component.
Matt (03:23):
There is, which gets us, I think grouped in with a lot of
AI companies.
Uh, but I, I would never haveconsidered us to be a real AI
company.
I don't like the term AI,artificial intelligence.
I think a better term would bemachine intelligence because
there's nothing artificial aboutintelligence.
It's either intelligent or it'snot.
Christy (03:42):
And in order to enable machine intelligence, we first
have to have an understanding ofwhat intelligence is.
And that's what leads us to thebrain because that's really the
best, if not the only examplethat we have.
Matt (03:54):
We run in a lot of neuroscience circles.
So it's sort of obvious that ifyou want to, to understand
intelligence, you know, the onlything in the universe we know of
that's intelligent is the brain.
So let's try and understandthat.
But I do think there's, there'sa lot of computer scientists and
mathematicians who have beenworking from the other direction
trying to crack that problem ofintelligence in lots of
(04:15):
different, other interestingways.
So it's just, that has neverbeen our mission.
It's not to build intelligentmachines.
It's to understand intelligence.
Christy (04:23):
Yes.
So yeah, so that, that's reallybeen our mission from the very
beginning.
And, and we've been at this formore than 10 years.
So let's talk a little bit aboutwhat we've learned so far along
the way about how the brain, howthe brain works.
Matt (04:39):
I think we've had a couple of big achievements and one of
them is about sequence memoryand in 2013 when we released all
of our, our core code as opensource.
That, that was the big discoverythat we had was about sequence
memory.
And it's really about how yourbrain memorizes spatial patterns
(05:01):
over time.
And the time thing is reallycore in that it's a, it's how
cells, how they, how theyconnect together to, to create
these sequences that can bereplayed, like a melody,
Christy (05:14):
Like a melody.
Yes.
And that's the example that Ilove, that works best for me
because when you think about howmany, how many melodies you
recognize, right?
Hundreds, at least, right?
If not, if not more.
And when you're listening to asong, you are constantly
predicting what, what the nextpart is, because you know the
(05:36):
song, you know what the nextnote will be, and if it's
different, you will detect thatas something different.
You will detect that as ananomaly.
Matt (05:44):
Right.
And you know, you, you couldeven invoke sequences in
people's brains listening to thepodcast right now.
So get this.
(Matt sings) Everybody who'sever heard that song that
classic, I'm assuming it'sBeethoven, I hope it's
Beethoven, I'm going to assume,I'm just going to call it
Beethoven.
But I'm not into classical musicreading neuroscience here.
(06:06):
So it doesn't matter if I'mwrong about about classical
music anyway.
Everyone knows that song.
And, and you're probably hearingit in your head right now.
We've invoked it just by thosefour notes, just by playing
those four notes in that order,at any instrument, in any voice,
you can sing those notes andinvoke that sequence of how you
experienced that song in yourlife in everybody's brain.
(06:28):
I think that's prettyinteresting.
But that idea, you know, thatsequence memory that ties those
notes together, the way yourbrain has memorized that over
time, that's at the core of thatfirst discovery, right?
We're talking about in theneuron paper.
We'll link to it in the shownotes if you want to read the
details.
Christy (06:44):
Yeah, so as Matt said, that that was really one of the
first big discoveries.
I think the second area ofdiscovery builds on that
sequence memory and is about howwe learn objects over time
through movement.
So when you think aboutlistening to a song, you can be
sitting absolutely still andtake in the song and you're
still predicting what note willbe next, but most of how we
(07:09):
experience the world is as aresult of our own movements.
So that's what our second kindof area for discovery starts to
address.
Matt (07:18):
Yeah, and so I think about, I like to think about
this new part of the theoryabout object representation and
about movement in sort of threedifferent ways.
There's a learning an object andthe best metaphor for this is if
you pretend to reach into a darkbox, you can't see anything in
it, but you just put your fingerin and you touch something in
the box.
I did this experiment with Jeffin the last episode, but at a
(07:40):
basic level, if you're feelingsomething that you've never felt
before, you've got to touch itall over with your hand.
You've got to traverse thatobject.
You, you've got to move acrossit to like build out a
representation of it in yourbrain and your imagination.
And that idea of movement overtime is core to how we build up
(08:01):
object representations.
It's through sequences, throughsequence memory, and that's why
we need the first part of thetheory that we just talked
about.
So it's crucial to have thefirst part for this object
recognition to build on top ofit.
Um, so you learn objects throughsequences and also you have to
be able to infer objects.
If you were to touch somethingin a box and say you've touched
(08:22):
a lot of things in your life,you could imagine what it might
be.
If it's furry, it could only bea certain amount of things.
So that's inference, you know,that's, that's like, I'm going
to guess what this is based oninterest based on what I know so
far.
Um, and then the third part isprediction.
Prediction also involvessequence memory because it, it
gives you the ability to predictan object state knowing that
(08:45):
it's been, you've seen it inthis sequence of states over
time you get an indication ofhow likely a state of that
object will be in the future.
Um, so I think those three partsare really core to this, uh,
this idea of, of objectrepresentations with sequence
memory.
So that's object recognition ina nutshell, I guess.
Christy (09:06):
So if you're interested in learning more about these
discoveries, we have papers andwe'll link to them in the show
notes.
The paper is pretty detailed,but we have some additional
supporting resources as well.
So I think a natural nextquestion would be, okay, so
that's what you've learned.
(09:26):
What about applications?
What have you built?
Matt (09:29):
Yeah.
Putting aside the whole objectrecognition thing, the, the, the
older stuff about sequencememory, and that was in the
"neuron paper," that's beenaround since 2013 in the public.
Um, it has some interestingapplications for anomaly
detection.
We really think this wholesequence memory was a
breakthrough and I think there'sa lot of still untapped
potential there.
(09:50):
There are some commercialapplications being built around
streaming anomaly detection.
I think there's otheropportunities for people to work
in that area, especially whatI'm still excited about is
geospatial, objects movingthrough time and space.
That's something that currentHTM techniques can take
(10:10):
advantage of right now.
So you can track things movingthrough space over time and, uh,
encode how anomalous thatobject's behavior is.
So, uh, if you're talking aboutlogistics or routing for, for
planes or trains or automobilesor, or whatever.
There's a lot of potentialapplications there in anomaly
(10:31):
detection.
Christy (10:32):
Yeah, it's really anything, anything that has data
streaming from something, right,whether it's a sensor or a GPS
signal, or medical device or Imean anything that has a
sequence of data where you canmake predictions, you know,
which is a lot of things,especially in the, the, you
(10:54):
know, Internet of things era.
Matt (10:56):
Yeah.
And a lot more things nowadayshave locations that move around
too.
Christy (11:01):
So we have a number of example applications that are
available for anyone to look atit, to play with.
(11:08):
Oh, HTM Studio, Talk about HTM Studio.
Yes.
So HTM studio is near and dearto my heart because it doesn't
require any technical skills.
So it's a tool that we releaseda couple of years ago that's
specifically designed to let youtest our technology on your data
to see if it finds interestinganomalies.
(11:29):
So anytime people come to us andthey think they have a use case
or a potential application andthey want to know if they can
use HTM, that's the first placewe send them because it's, you
know, it's a minimal investment,it's easy.
It's a good way to test it.
Matt (11:46):
All you have to do is get your data in a CSV format.
That's pretty simple.
And uh, and it takes care of therest, honestly.
A lot of people will come intothe technical forum on, you
know, the hackers forum in our,in our community and ask,"Is
this HTM going to work well onthis data and I'll just tell
them, go download HTM Studio.
It's just, uh, it, it works onMac and windows, so you can
(12:08):
install it easily and then justupload a CSV into it.
It's got sample applications.
It'll run NuPIC in thebackground.
It runs HTM in the backgroundand it'll tell you where the
anomalies are and that streamingdata.
So if you have any of that datathat Christy was just talking
about, you know, from devices orwhatever, give it a shot because
if it's, if it has patterns overtime, hourly patterns, daily
(12:30):
patterns or even weekly andmonthly patterns, it'll find it
given enough data.
Christy (12:33):
That's really what makes this approach to anomaly
detection unique is that it'snot just about finding a spike
or using a threshold or youknow, kind of those more
traditional techniques.
It's really able to find themore subtle,
(12:47):
the nuanced stuff.
Yes.
So all of our exampleapplications are available for
you to see, the code isavailable in open source, which
actually brings us to our nextthing you need to know about
Numenta.
Matt (12:59):
Yeah.
And that's that we are verytransparent in everything that
we do.
We believe in open science andso if you follow Subutai, our VP
of research on Twitter forexample, you'll know he's always
posting about open scienceissues.
We only publish in open journalsthat are free and open access.
So we, we really encourage that.
(13:19):
We put as much as we can online,you know, a lot of times I'll
record meetings and put them onYouTube, or if someone on a
forum is asking about something,I'll just grab someone and, and
we'll answer the question andthrow it out there.
Christy (13:33):
Yeah, the daily research code, I mean it's, it's
all out there, right?
It's all accessible.
There's nothing really hidden.
Matt (13:39):
No, all of our research papers are online and free, like
I said.
We're trying to get more of theresearch paper code in a
standardized format so that it'sreally easy to run for people
and to replicate all of ourexperiments.
Christy has a great events pageon numenta.com that's got all of
our upcoming talks andconferences, including previous
(14:02):
talks and conferences and shealways put slides and videos in
there.
So there's a lot of media thatyou can consume from our
previous events and our eventspage that's really nice.
Christy (14:13):
We also have an educational youtube series
called HTM School, which Matthosts and you want to talk about
HTM School?
Matt (14:19):
It's a from the ground up explanation of HTM theory.
So we really start from thebeginning, assuming you know
nothing about computer science,about neuroscience and, and
really attempt to build up thetheory in a way that makes sense
from scratch.
Uh, and right now it's up to 14episodes or something from, from
binary encodings and semanticsall the way up to grid cells and
(14:44):
columns and layers and all ofthat stuff.
So it's, uh, it's unfinished.
I know there's probably going tobe a few more episodes at some
point, but it's a greateducational resource if you
don't feel like reading thepapers, you can get a full
understanding, I think just bywatching these videos.
You may have to watch a fewtimes because I packed them full
of information, but you can getthe concepts of HTM theory down
(15:08):
just by watching these videos
Christy (15:09):
and each one is about 15 minutes long.
Right?
So it's
Matt (15:12):
give or take the bloopers and outtakes.
(15:16):
The outtakes, my favorite.
I'm also putting together morevisual documentation.
So we're continuing to work ondocs to make the theory even
more approachable.
Christy (15:24):
So that really brings us to our last point, which is
how you can get involved.
You know, we'd love for as manypeople to get involved as
possible and we also understandthat there are different types
of people out there, right?
Not everyone's a neuroscientist,not everyone's going to read
every peer reviewed journalpaper, that we release, and we
understand that.
So if you're looking for more ofthe, you know, 30,000 foot view,
(15:48):
maybe you're interested in AIand you're interested in brain
related approaches.
We have a few pieces and we'llput these in the show notes.
We'll put links in the shownotes and a few pieces that Jeff
has written about this topic.
One called The Secret to StrongAI, one called"The Thousand
Brains Model of Intelligence"and an article he wrote for a
(16:10):
special edition of IEEE Spectrumthat talks about what
intelligent machines need tolearn from the neocortex.
And all three of these articlesreally highlight our most recent
advances in the research andtalk about why we think they're
so important for and theimplications for machine
intelligence.
Matt (16:30):
Yeah, they, they try and explain what we're doing and why
it's important sort of indifferent ways to different
audiences.
Christy (16:35):
Right.
And then of course, if you dowant to get down into the
scientific details and thenhopefully you've already
listened to the first coupleepisodes of this podcast, but we
have all of our papers availableon the website as well.
And for each of our peerreviewed papers, we have an FAQ
section that talks about kind ofthe highlights and what you need
(16:57):
to know about each paper.
Matt (16:59):
And if passive learning is not enough for you, you should
join our online community orwe've got a really robust forum
called HTM forum.
There's not only a hackercommunity of, of people that are
trying to build things with HTMsor build their own HTMs and
code, uh, but there's a, there'sa thriving theory community as
well that are interested in therepercussions of HTM theory
(17:22):
extensions to HTM theory,linking it to other parts of
neuroscience or psychology.
There's lots of people with lotsof interesting ideas on our
public forums.
So join there if you want tolearn more and really interact
with our community.
Yeah.
Christy (17:34):
And Matt manages that community, so he sees all the
posts that come in and makessure things get answered.
(17:40):
I do,yes.
You'll have a warm welcome.
It is a friendly community.
I think that's fair to say.
And it's also a place that, thatalmost becomes a bit of a, at
times a technical support forumwhere people who have questions
about HTM or Numenta can, can goto the forum and some of the
community members can answer.
Matt (17:59):
You know, our, our community has gotten mature
enough now that I've got a smallgroup of people that I can
depend upon to answer questionswhen new people come in and
they, and they know the theorywell enough that they can answer
the questions and that's reallynice to get the community to
that maturity level.
Christy (18:16):
And you can also join without posting.
Matt (18:19):
Oh absolutely.
(18:19):
If you just want to read the threads You can lurk.
You don't even have to to join,you know, you can just go
through and read all themessages.
But it's more fun if you, if youmake an account because then you
can like posts and interact withpeople.
Christy (18:33):
So hopefully we'll see you there.
(18:35):
I hope so.
So that's a broad summary ofNumenta.
We hope you enjoyed that, youknow, five things that you
should know.
What were they, Christy?
We should review them.
Christy (18:44):
Yes.
So the first was(1) our mission,our scientific mission to
understand the brain,
Matt (18:50):
(2) what have we learned so far, and(3) some things about
our anomaly detectionapplications,
Christy (18:54):
(4) Our commitment to open access and research
transparency,
Matt (18:59):
(5) and all the ways that you can get more involved in our
community or learn more aboutwhat we do.
(19:07):
Yes.
So there you have it.
Hopefully you found that helpfuland hopefully you feel like you
know us a little bit more.
It's been a pleasure sharing allthis information with you,
podcast audience.
Please keep listening, subscribeto our podcast Numenta On
Intelligence.
Thanks.
Matt (19:26):
Thanks for listening to Numenta On Intelligence.
To learn more about Numenta andthe progress we're making on
understanding how the brainworks, go to numenta.com.
Hi, this is Christy Maver.
(00:02):
And I'm Matt Taylor and you're listening to the Numenta On
Intelligence podcast.
Today's episode is designed forpeople who might be unfamiliar
with Numenta and our work.
Hopefully this episode willpique your interest and entice
you to follow along with whatwe're doing.
If you want to stay up to date,the best way to do that is to
sign up for our newsletter.
(00:25):
You'll find a link to subscribeon our website at numenta.com.
We typically send a newsletterevery one to two months.
Basically, anytime we have newsto share about upcoming events,
new papers we've published, andpartner updates, just to name a
few.
Subscribe today to make sure youdon't miss any Numenta news.
(00:45):
Thanks for joining us.
Hi Christy.
Hi Matt.
How are you?
I'm doing great.
I was thinking about the lastepisode, which was episode one
we did with Jeff Jeff Hawkins,our co-founder.
Two parts.
We're really bad at numberingepisodes, so we have episode
zero, episode one, part one, andepisode one, part two, and now
this is episode two.
This is episode two slash three.
(01:07):
Makes total sense.
Anyway, that last episode wentreally deep and we just want to
let everybody know that it's notalways going to be like that.
Right.
So Matt and I were thinkingafter listening to the last
episode, Let's do an episodethat's like the opposite, not a
huge deep dive, just a shallowskim on the surface of what it
is.
So maybe if you're new toNumenta or a beginner in brains
(01:32):
or just mildly curious, this isa great episode to start on for
you.
Especially if the last one justwent way over your head.
Don't give up on the podcast.
We're going to mix it up.
We're going to have someinteresting interviews with
neuroscientists and interns andceos coming up.
So what we want to do in thisepisode is we want to break down
the top five things you need toknow about Numenta,
Matt (01:54):
The first one being what is our mission?
Why do we exist, what are wehere for?
(02:00):
What are we here for Matt?
Well, I'll tell you what we'rehere for.
When I first heard about Numentais right after I read the book
On Intelligence by Jeff and thatreally affected the way I
thought.
So I looked it up.
I looked at what Jeff Hawkinswas doing and I saw that he had
founded this company calledNumenta and their mission was to
understand intelligence and Ithought that was just an amazing
(02:23):
mission for a company in SiliconValley to have and that is still
our mission even now over 10years later, is to understand
how intelligence works in thebrain.
Christy (02:32):
Yeah.
And you know, we often talkabout how we actually have this
dual mission of number one tounderstand how the brain works,
understand intelligence, andthat is very much a scientific
mission.
Brain theory driven, deepneuroscience details.
Matt (02:50):
That is number one by Jeff because he wants to know, like
honestly because he's curious,right?
Christy (03:00):
I think more than just Jeff wants to know.
Matt (03:01):
Yeah, I mean it's, there's a lot of passion behind our
mission because another thing Ilike about Numenta.
Christy (03:06):
So the first part of the mission is a scientific one.
There is a secondary mission,which is to apply those
principles of intelligence tosoftware to enable intelligent
machines.
So there is a machineintelligence, AI component.
Matt (03:23):
There is, which gets us, I think grouped in with a lot of
AI companies.
Uh, but I, I would never haveconsidered us to be a real AI
company.
I don't like the term AI,artificial intelligence.
I think a better term would bemachine intelligence because
there's nothing artificial aboutintelligence.
It's either intelligent or it'snot.
Christy (03:42):
And in order to enable machine intelligence, we first
have to have an understanding ofwhat intelligence is.
And that's what leads us to thebrain because that's really the
best, if not the only examplethat we have.
Matt (03:54):
We run in a lot of neuroscience circles.
So it's sort of obvious that ifyou want to, to understand
intelligence, you know, the onlything in the universe we know of
that's intelligent is the brain.
So let's try and understandthat.
But I do think there's, there'sa lot of computer scientists and
mathematicians who have beenworking from the other direction
trying to crack that problem ofintelligence in lots of
(04:15):
different, other interestingways.
So it's just, that has neverbeen our mission.
It's not to build intelligentmachines.
It's to understand intelligence.
Christy (04:23):
Yes.
So yeah, so that, that's reallybeen our mission from the very
beginning.
And, and we've been at this formore than 10 years.
So let's talk a little bit aboutwhat we've learned so far along
the way about how the brain, howthe brain works.
Matt (04:39):
I think we've had a couple of big achievements and one of
them is about sequence memoryand in 2013 when we released all
of our, our core code as opensource.
That, that was the big discoverythat we had was about sequence
memory.
And it's really about how yourbrain memorizes spatial patterns
(05:01):
over time.
And the time thing is reallycore in that it's a, it's how
cells, how they, how theyconnect together to, to create
these sequences that can bereplayed, like a melody,
Christy (05:14):
Like a melody.
Yes.
And that's the example that Ilove, that works best for me
because when you think about howmany, how many melodies you
recognize, right?
Hundreds, at least, right?
If not, if not more.
And when you're listening to asong, you are constantly
predicting what, what the nextpart is, because you know the
(05:36):
song, you know what the nextnote will be, and if it's
different, you will detect thatas something different.
You will detect that as ananomaly.
Matt (05:44):
Right.
And you know, you, you couldeven invoke sequences in
people's brains listening to thepodcast right now.
So get this.
(Matt sings) Everybody who'sever heard that song that
classic, I'm assuming it'sBeethoven, I hope it's
Beethoven, I'm going to assume,I'm just going to call it
Beethoven.
But I'm not into classical musicreading neuroscience here.
(06:06):
So it doesn't matter if I'mwrong about about classical
music anyway.
Everyone knows that song.
And, and you're probably hearingit in your head right now.
We've invoked it just by thosefour notes, just by playing
those four notes in that order,at any instrument, in any voice,
you can sing those notes andinvoke that sequence of how you
experienced that song in yourlife in everybody's brain.
(06:28):
I think that's prettyinteresting.
But that idea, you know, thatsequence memory that ties those
notes together, the way yourbrain has memorized that over
time, that's at the core of thatfirst discovery, right?
We're talking about in theneuron paper.
We'll link to it in the shownotes if you want to read the
details.
Christy (06:44):
Yeah, so as Matt said, that that was really one of the
first big discoveries.
I think the second area ofdiscovery builds on that
sequence memory and is about howwe learn objects over time
through movement.
So when you think aboutlistening to a song, you can be
sitting absolutely still andtake in the song and you're
still predicting what note willbe next, but most of how we
(07:09):
experience the world is as aresult of our own movements.
So that's what our second kindof area for discovery starts to
address.
Matt (07:18):
Yeah, and so I think about, I like to think about
this new part of the theoryabout object representation and
about movement in sort of threedifferent ways.
There's a learning an object andthe best metaphor for this is if
you pretend to reach into a darkbox, you can't see anything in
it, but you just put your fingerin and you touch something in
the box.
I did this experiment with Jeffin the last episode, but at a
(07:40):
basic level, if you're feelingsomething that you've never felt
before, you've got to touch itall over with your hand.
You've got to traverse thatobject.
You, you've got to move acrossit to like build out a
representation of it in yourbrain and your imagination.
And that idea of movement overtime is core to how we build up
(08:01):
object representations.
It's through sequences, throughsequence memory, and that's why
we need the first part of thetheory that we just talked
about.
So it's crucial to have thefirst part for this object
recognition to build on top ofit.
Um, so you learn objects throughsequences and also you have to
be able to infer objects.
If you were to touch somethingin a box and say you've touched
(08:22):
a lot of things in your life,you could imagine what it might
be.
If it's furry, it could only bea certain amount of things.
So that's inference, you know,that's, that's like, I'm going
to guess what this is based oninterest based on what I know so
far.
Um, and then the third part isprediction.
Prediction also involvessequence memory because it, it
gives you the ability to predictan object state knowing that
(08:45):
it's been, you've seen it inthis sequence of states over
time you get an indication ofhow likely a state of that
object will be in the future.
Um, so I think those three partsare really core to this, uh,
this idea of, of objectrepresentations with sequence
memory.
So that's object recognition ina nutshell, I guess.
Christy (09:06):
So if you're interested in learning more about these
discoveries, we have papers andwe'll link to them in the show
notes.
The paper is pretty detailed,but we have some additional
supporting resources as well.
So I think a natural nextquestion would be, okay, so
that's what you've learned.
(09:26):
What about applications?
What have you built?
Matt (09:29):
Yeah.
Putting aside the whole objectrecognition thing, the, the, the
older stuff about sequencememory, and that was in the
"neuron paper," that's beenaround since 2013 in the public.
Um, it has some interestingapplications for anomaly
detection.
We really think this wholesequence memory was a
breakthrough and I think there'sa lot of still untapped
potential there.
(09:50):
There are some commercialapplications being built around
streaming anomaly detection.
I think there's otheropportunities for people to work
in that area, especially whatI'm still excited about is
geospatial, objects movingthrough time and space.
That's something that currentHTM techniques can take
(10:10):
advantage of right now.
So you can track things movingthrough space over time and, uh,
encode how anomalous thatobject's behavior is.
So, uh, if you're talking aboutlogistics or routing for, for
planes or trains or automobilesor, or whatever.
There's a lot of potentialapplications there in anomaly
(10:31):
detection.
Christy (10:32):
Yeah, it's really anything, anything that has data
streaming from something, right,whether it's a sensor or a GPS
signal, or medical device or Imean anything that has a
sequence of data where you canmake predictions, you know,
which is a lot of things,especially in the, the, you
(10:54):
know, Internet of things era.
Matt (10:56):
Yeah.
And a lot more things nowadayshave locations that move around
too.
Christy (11:01):
So we have a number of example applications that are
available for anyone to look atit, to play with.
(11:08):
Oh, HTM Studio, Talk about HTM Studio.
Yes.
So HTM studio is near and dearto my heart because it doesn't
require any technical skills.
So it's a tool that we releaseda couple of years ago that's
specifically designed to let youtest our technology on your data
to see if it finds interestinganomalies.
(11:29):
So anytime people come to us andthey think they have a use case
or a potential application andthey want to know if they can
use HTM, that's the first placewe send them because it's, you
know, it's a minimal investment,it's easy.
It's a good way to test it.
Matt (11:46):
All you have to do is get your data in a CSV format.
That's pretty simple.
And uh, and it takes care of therest, honestly.
A lot of people will come intothe technical forum on, you
know, the hackers forum in our,in our community and ask,"Is
this HTM going to work well onthis data and I'll just tell
them, go download HTM Studio.
It's just, uh, it, it works onMac and windows, so you can
(12:08):
install it easily and then justupload a CSV into it.
It's got sample applications.
It'll run NuPIC in thebackground.
It runs HTM in the backgroundand it'll tell you where the
anomalies are and that streamingdata.
So if you have any of that datathat Christy was just talking
about, you know, from devices orwhatever, give it a shot because
if it's, if it has patterns overtime, hourly patterns, daily
(12:30):
patterns or even weekly andmonthly patterns, it'll find it
given enough data.
Christy (12:33):
That's really what makes this approach to anomaly
detection unique is that it'snot just about finding a spike
or using a threshold or youknow, kind of those more
traditional techniques.
It's really able to find themore subtle,
(12:47):
the nuanced stuff.
Yes.
So all of our exampleapplications are available for
you to see, the code isavailable in open source, which
actually brings us to our nextthing you need to know about
Numenta.
Matt (12:59):
Yeah.
And that's that we are verytransparent in everything that
we do.
We believe in open science andso if you follow Subutai, our VP
of research on Twitter forexample, you'll know he's always
posting about open scienceissues.
We only publish in open journalsthat are free and open access.
So we, we really encourage that.
(13:19):
We put as much as we can online,you know, a lot of times I'll
record meetings and put them onYouTube, or if someone on a
forum is asking about something,I'll just grab someone and, and
we'll answer the question andthrow it out there.
Christy (13:33):
Yeah, the daily research code, I mean it's, it's
all out there, right?
It's all accessible.
There's nothing really hidden.
Matt (13:39):
No, all of our research papers are online and free, like
I said.
We're trying to get more of theresearch paper code in a
standardized format so that it'sreally easy to run for people
and to replicate all of ourexperiments.
Christy has a great events pageon numenta.com that's got all of
our upcoming talks andconferences, including previous
(14:02):
talks and conferences and shealways put slides and videos in
there.
So there's a lot of media thatyou can consume from our
previous events and our eventspage that's really nice.
Christy (14:13):
We also have an educational youtube series
called HTM School, which Matthosts and you want to talk about
HTM School?
Matt (14:19):
It's a from the ground up explanation of HTM theory.
So we really start from thebeginning, assuming you know
nothing about computer science,about neuroscience and, and
really attempt to build up thetheory in a way that makes sense
from scratch.
Uh, and right now it's up to 14episodes or something from, from
binary encodings and semanticsall the way up to grid cells and
(14:44):
columns and layers and all ofthat stuff.
So it's, uh, it's unfinished.
I know there's probably going tobe a few more episodes at some
point, but it's a greateducational resource if you
don't feel like reading thepapers, you can get a full
understanding, I think just bywatching these videos.
You may have to watch a fewtimes because I packed them full
of information, but you can getthe concepts of HTM theory down
(15:08):
just by watching these videos
Christy (15:09):
and each one is about 15 minutes long.
Right?
So it's
Matt (15:12):
give or take the bloopers and outtakes.
(15:16):
The outtakes, my favorite.
I'm also putting together morevisual documentation.
So we're continuing to work ondocs to make the theory even
more approachable.
Christy (15:24):
So that really brings us to our last point, which is
how you can get involved.
You know, we'd love for as manypeople to get involved as
possible and we also understandthat there are different types
of people out there, right?
Not everyone's a neuroscientist,not everyone's going to read
every peer reviewed journalpaper, that we release, and we
understand that.
So if you're looking for more ofthe, you know, 30,000 foot view,
(15:48):
maybe you're interested in AIand you're interested in brain
related approaches.
We have a few pieces and we'llput these in the show notes.
We'll put links in the shownotes and a few pieces that Jeff
has written about this topic.
One called The Secret to StrongAI, one called"The Thousand
Brains Model of Intelligence"and an article he wrote for a
(16:10):
special edition of IEEE Spectrumthat talks about what
intelligent machines need tolearn from the neocortex.
And all three of these articlesreally highlight our most recent
advances in the research andtalk about why we think they're
so important for and theimplications for machine
intelligence.
Matt (16:30):
Yeah, they, they try and explain what we're doing and why
it's important sort of indifferent ways to different
audiences.
Christy (16:35):
Right.
And then of course, if you dowant to get down into the
scientific details and thenhopefully you've already
listened to the first coupleepisodes of this podcast, but we
have all of our papers availableon the website as well.
And for each of our peerreviewed papers, we have an FAQ
section that talks about kind ofthe highlights and what you need
(16:57):
to know about each paper.
Matt (16:59):
And if passive learning is not enough for you, you should
join our online community orwe've got a really robust forum
called HTM forum.
There's not only a hackercommunity of, of people that are
trying to build things with HTMsor build their own HTMs and
code, uh, but there's a, there'sa thriving theory community as
well that are interested in therepercussions of HTM theory
(17:22):
extensions to HTM theory,linking it to other parts of
neuroscience or psychology.
There's lots of people with lotsof interesting ideas on our
public forums.
So join there if you want tolearn more and really interact
with our community.
Yeah.
Christy (17:34):
And Matt manages that community, so he sees all the
posts that come in and makessure things get answered.
(17:40):
I do,yes.
You'll have a warm welcome.
It is a friendly community.
I think that's fair to say.
And it's also a place that, thatalmost becomes a bit of a, at
times a technical support forumwhere people who have questions
about HTM or Numenta can, can goto the forum and some of the
community members can answer.
Matt (17:59):
You know, our, our community has gotten mature
enough now that I've got a smallgroup of people that I can
depend upon to answer questionswhen new people come in and
they, and they know the theorywell enough that they can answer
the questions and that's reallynice to get the community to
that maturity level.
Christy (18:16):
And you can also join without posting.
Matt (18:19):
Oh absolutely.
(18:19):
If you just want to read the threads You can lurk.
You don't even have to to join,you know, you can just go
through and read all themessages.
But it's more fun if you, if youmake an account because then you
can like posts and interact withpeople.
Christy (18:33):
So hopefully we'll see you there.
(18:35):
I hope so.
So that's a broad summary ofNumenta.
We hope you enjoyed that, youknow, five things that you
should know.
What were they, Christy?
We should review them.
Christy (18:44):
Yes.
So the first was(1) our mission,our scientific mission to
understand the brain,
Matt (18:50):
(2) what have we learned so far, and(3) some things about
our anomaly detectionapplications,
Christy (18:54):
(4) Our commitment to open access and research
transparency,
Matt (18:59):
(5) and all the ways that you can get more involved in our
community or learn more aboutwhat we do.
(19:07):
Yes.
So there you have it.
Hopefully you found that helpfuland hopefully you feel like you
know us a little bit more.
It's been a pleasure sharing allthis information with you,
podcast audience.
Please keep listening, subscribeto our podcast Numenta On
Intelligence.
Thanks.
Matt (19:26):
Thanks for listening to Numenta On Intelligence.
To learn more about Numenta andthe progress we're making on
understanding how the brainworks, go to numenta.com.
Popular Podcasts
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Intentionally Disturbing
Join me on this podcast as I navigate the murky waters of human behavior, current events, and personal anecdotes through in-depth interviews with incredible people—all served with a generous helping of sarcasm and satire. After years as a forensic and clinical psychologist, I offer a unique interview style and a low tolerance for bullshit, quickly steering conversations toward depth and darkness. I honor the seriousness while also appreciating wit. I’m your guide through the twisted labyrinth of the human psyche, armed with dark humor and biting wit.