Bestand wählen
Merken

Kickoff (2015 Spring NuPIC Hackathon)

Zitierlink des Filmsegments
Embed Code

Automatisierte Medienanalyse

Beta
Erkannte Entitäten
Sprachtranskript
but at the moment we are using a lot of those things are right that the higher the
1st deflector where you get the idea of this 1st the 1st half of the reporting generally back there so that other places so when were doing sessions during the quarter we will again be here might be interested in what we're doing so you might need quality is that if you would like to take any photographs or videos you've read this is a variable transparent thing we're doing here so if you want use the hashtag social media that you the thanks to everyone who has helped you this on request and Jerry Wright the there at the registration desk downstairs and instrumental in for a logistics and planning and think you just and then the rest of the big engineers on the right here the here willing to help you if you can solve friends are about what you can do with your data we will have a really about research so this is among the 3 of us westerns and they can when someone can also thinks point out that the is students while they were also anyway forget over that let us use their space I just would you out and as a couple of things to hackers motivated and you find I is the title of the 1st few minutes you will know what the have is great like what they can do and I think of you know it when they that hill coming here and I would like to hear 1st time I saw it in the in and so it's so you experience for us so we have some form of social integration and that which I think is really about you guys with the life personal thanks for coming here you know what we're working on that question what you want to work on the wall I think is extremely extremely important work is not just the technology will be for humans in many ways we are talking about the which 1 reverse-engineer the neocortex with partners and we want people to do with the principles of because all of the open source component of and and I think the implications of this work are very very profound in the future will get a chance to do something that is not the guy fugal layout of right what my friends on this kind of method time right now but it's it's it's something it's it's really happening is starting to form the and is the center of the next year many decades of genes in the core principles and you comes do so if they have other people to participate and that is when you come along with me here so we can make a list of all of you support work and I don't really want to live in much all at once thank you come in order to try to make a successful and this event in the whole and and you know the whole thing here and so on and they come in and I'll be around here myself off for a part of data about the moral and so if you want something you want and that was not really in charge of running everything and they work like major vintage activities you know it's and I like this is the 1st work that OK USA boring work also in general so 1st of resources for you everything that you need to know what should be a vector of words such that you you go there that makes to are scheduled to go to product to resources to everything we were interested in all that stuff and we will register on the website called challenge posed a challenge because not some if you're going to have to have them at the end of this you need to register their more influence on the work sort all of slides also so use either 1 2 links and that might be useful to you you can also find the unit of selection at work so the schedule I'm doing the around in 30 years that this would have a couple of engineers and show how you samples of things they've done with it using the chain is as a demo using the unity-gain engine and the data from that interview and Austin as something with the object tango such interest in 11 30 to 12 30 where we will be available probably and there's a big conference room in the back here so we will have to big engineers to help you get to be installed if you're having problems getting compiled and running on your laptop that's well 30 to 1 30 in the room that behind the projection screen we're going to have to really brainstorming session on of music theory with new picture of the idea behind this is to try to help with ideas on how to encode the data into sparse representations so that so that we can use our understanding user name this is really interesting we have a special guest with us that's going to help with the user these are in-house using Eurostat Charlie Gillingham from planning grows is the keyboard is a growth and there is also a computer scientist you know that also will will be running at such an overly biggest brains of music theory and how we can encode data into sports representations but and I'm going to do the work session on the state of you need to talk about 230 here the data you want to go around like with communities like what's going on in the community and what kind of projects are happening where we're going what with the
published over the past year so it's not focus on people interested in contributing oriented development progress what happened with the new project itself and we also have a higher risk of model from open BCI who brought a bunch of hardware stuff open BCI because of the why why I she gets so that we get those things going ahead with contacts and get g information from the brain and we got a couple that's in the past getting that data the take and using it to control things so that could be interesting if you're interested in that we use it in that session I will probably be back in this room behind the tomorrow and had nothing on the agenda is done with the war so if you're going to get you ready to work buildings about the facilities in the vertices and this is the sign and the leave 1 exit in this room suspected and all the words in here but don't open pulling pushing the all sliding this is that problem in the past half of this year where that no way is you're doing it wrong slide toward at work and I mentioned that was earlier admitted women have different so the only 1 where have just take a look are white writers who will be asked very fast and it's free and about this year's sleeping here feel free to write we want the lowers the light something because there's no doubt of but if you want to talk about accommodations you can talk to me later and the reason somewhere perhaps a little more protocol UGG that is to get in the door tomorrow this thing and bring back tomorrow on the back of it is the real important things is what the security guard along with you so keep that also has a part or usually provide food these events because of logistics problems this facility where people to to provide meals so we we compensated by by giving you if it if you can't find food out on the street there's something wrong with there is time to move around here it's a lot of ways to so please don't use that word so that you can really but there is transcendental in all ways use and I will be making sweets every few hours and so the the traffic lines here you just read the respectful of the facility you don't know what you should be touching something common asked me utterly you can touch and also please take a look at the code of conduct is typical of the kind of don't do anything that's that's harassing or offensive people that you think it might even say that probably is expensive so don't come up with any that might offend anyone you want specific details read here OK but to me is very obviously we're all the same room so that's 1 of the things that you have a technical questions that here can answer or you find an answer to a share with people we have mailing lists you can go to the and works like there is a general discussion questions probably that has 1 were developed 1 for you so feel free to sign up for that and and remember from the leading edges of the slides the slides are available and the works such all these things but we also have a chat room on tyrosine which I sometimes on and we also have a bigger and you have you have account that's the 1 that generally has the most attractive and spread the word would have said that it looks like it's vanity thing a going into areas it's social media systems OK you questions for each 1 of these raise your hand if you've ever been to act on board lot of people that's the uh that the who has actually filed and build on their laptop better indication of how busy we're going to be helping to maybe have a set of I who is running in the example code tutorials that like that of media and who has like taking their own data were manipulated their own something that overall cascade is that some of the school they missed so some more resources for you for those of you did not raise your hand and fundamental that workers are link to the wiki and I would say this probably the best that were looking at getting started to do with the wiki is laid out a way to kind of direct you have what you build and install you tutorials on how to use new and how to learn about what it is doing in the theory behind it that sort of thing so that that's a good resource to start at the URIs suggested links especially for for you hackers you're new to you and you want to figure out what it's all about and we've got a couple of tutorials people generally about the energy consumption over time this really nice use case for new data because it recognizes patterns of energy consumption in buildings really well so both of these screencasts in the sample code in me talking on on 1 of our model also the avoidance understandable there's 1 for prediction and there's 1 for anomaly detection which follows we also have a beginner's guide to new which is also pretty in the previous and which is a really nice dive into the details of the coders how how we include scalar data for example into sports representations and so that's a good resource and there is a link here if you want to go examples on a weekly CPU
usage of spatial and temporal and classification what's so we don't we have good examples of all the stuff so if you see something that sparks interest indicate that might
fit in that starts with that and you know 1 of the variables in the also have a bunch of old emotional ones about from
all at once that they 14 that has lasted or something like that from the 1 before so you want some ideas of what other people in and try all the ones that are there if you do not to use your readers out of my numbers don't alright and so other each unit conditions you potentially use we have extended job which is the work of the new data point and each team algorithms by David there a direct embarrassed and the handling of water that's to introduce this year which is a quota implementation needs to it doesn't so closely followed are applied on the addition of these 2 problems like this and out of that it's going to OK look I studied the most the most important thing is if you have an idea about something to think about you think about you have a lot of data is the world's biggest question this is the data you're thinking about using perhaps world is that every time 1 of them is that it does this gestational you might have much success a key thing about HTML all optical world memories of world point that so make sure you have a problem on data and you can see how you it all or on there enough you know you have to learn the patterns from you know annual data points and this you have what's the what's in your there and the end of the smaller questionnaires answers yet it was let's let's have had no relevant questions presentation of what I talked about was the that and patterns in the face of the earth is that you you you knew his thousand points in OK a great we were the theory is building on your favorite programming like this that's great I love you will do this this that is just standing here telling people what they already know and here's here's my mind each year in whatever so please don't and that you do it right here that 1 but don't make that will be happy to help you do that this something
OK so was what we think about anyone having any ideas or right so I thought of music so I'm I'm really interested in the encoding mediated somehow so that you can understand that and this is an example that just uses a lot of talking about bringing about it here about the higher involves and and we actually had done in the past the 1st set actually published encoding of circumstance we song into scalar values investments we have reduced the entire so after about 20 to 24 year conditions it predicted levels on the so that was addressed to the but it wasn't like a true encoding of because they happen in the scaling here so that the real challenge here is how to get the data or anything of that musical representation into a format that includes more information than just the fish right there there's more to use it just it just note duration of that something I'm interested in of work that you're interested in that country music theory says we're having today because that's what we talking about but I wonder whether it is you who has an idea of something they want to really love to hear about it as you might motivate the people to help you implement the idea of artists building teams so I know the hazards of this region and you haven't you're willing to share with OK you're close come here the mine this town or as well and my idea was to use nuclear war last chance the Union's common recognition you like when you look at the absorption lines they look suspiciously close to source data representations for certain applications and because of that and I basically you you like something and then and whatever like doesn't have that tells you what's in the article that you should be more like that or the the latest passed from the I like the way but I can't think of a way to make temporal but it doesn't seem like something that will depend on how long you should write that because it's an maybe maybe there is a there is a video component it anyway and they don't have a security over time that maybe something to protect as long as you could somehow
reconstruct those into some some action yeah you don't get you so I want to do is is this on the by the way we do have a direct how someone what engineers project so that's something we don't think about going to yes you you you you you depend on what kind of data these words here and that is really interesting result of being in the temple wants you to use of semantic encodings during much of something that the words then sailed sentence right up in there but it's still even just the SDR component you might be useful for you in some way I don't know what data that there are some situations you can do a of it is that what we do with spatial patterns of research on this problem life's 1st this coming out of this actually have spent the I'm sure many exaggerated into the last time that it has to rely on that and reading about New Britain that had been thinking about how much you just like you have but in recent years in the center the which was trying to break the outcome of the buildings and there are 6 or 7 I am so all the algorithms and predicting with all of these models is that there's a lot money for ingredients in baseball season and it would that it always did it with high rate outcome for each gene is already it's really about thinking that that's really helpful and see if you get any easy bridges just by this is very significant events within engineer I'm I'm sure this is the very idea but I'm interested in the environment and so can we take some amount of data and use that to do something like that in turn off the lights for which or decrease and increase the temperature of rules based on if people around so that it was interested in something that's very there are lots of things I would have so we have various something with the New York Times data that's where the energy of that the alliance the relation with the following the building of everything's instrument to building you should be able to predict when people are going to be and what remains when things we use have money exchange and I'm working for a lot of your project for the York-Presbyterian Hospital this summer on regional morning and and also means here also structure of some and incorporate some kind of benefits you analysis before they get lost so that what makes people think which hospitals choose and words that students and natural processes in this project but hopefully there some applications in as we can help you my name's david true I had an idea for using Nietzscheans through and look at what motivated so can we add just a few extra sensors to car and get a better idea of what's wrong with the car for consumer applications or and get a better idea of how the drivers doing for like performance driving applications like it seems like there should be more information here then just OK this sensors to high so that engines do not like instead of the rumble came from the front right corner of the positive you know the way it was since it means that building that can help me die password regular applications you just don't have any data do it makes some those other validation application that it is reasonable so if you want to use the right 1 and had a cry and I am interested in trying to model student here I have some generated in the stands to the good and bad and thinking about making it and worldwide encoding the student and the question and maybe taxonomies about the question and then followed by the end of the year and just a simple you want the sequence but in order to get to world not to think about that support I have the building and then my dissertation behind model of the interaction between the campus in the cortex and visiting here brand experiments on that of lot of people that have sort of try to put into the the guy would win and that has to do with that of a rat will reject other it became this trajectory that we're going and you if you were in the trajectories generated by around or stimulates of the question is is how can work period time can our algorithm predicts which trajectories result work that's 1 time you have 1 work that's so consider using anomaly detection for a private action in the mobile space available advertisers punches so we are left to sort of set up their own the romantic using mostly the thresholds and you can use all the words like the idea of a country may be their locality and operating system and you could potentially use useful thing detect the types of trafficking German anomalous things that is to the security people working on up I think you the mobile space it a lot of a lot of publishers and use of their own and useful to consolidate creates some services and I think that about there are a lot of inclusions right ejection there goes along knowledge action greater rejection of the worker and I am mismatching of to coupling to go there was a question about this on the internet and this was some kind of this line so this means to figure out what all let's try to create a world war work 2 do centers of high if you train network is used to measure the amount that's produced per cent of that 1 bunch of Shakespeare's as well as Wikipedia was current sources during the prewhitened stimuli in the in the greatest amount so you know you could have well I guess you can network is keeping now up you can have versus the same thing my grows on you know as part of trying to understand the music algorithm and it was not interested in doing this like a sensitivity
assay of the algorithm itself using the generated generated data training on the classification problem and then perturb the data and say OK where classification start breaking down the randomness something to do with this so that you the video to compare the performance of the hybrid version the what it the I call on our understanding of its use this bring something about the universe and we use it to something very simple it's important something resembling the notion also went to point points where it's you know or or actually works so that the use of a room for instances the miners literary work in time series is a lot of times when the conviction of integers and video was to test the performance of this year the the highest nonstationary models show up on which each year at the time of duration of time series on the the of the partners she's college can recognized utterances and the subject of section 1 I am not want to have to with this so that the the interests of various articles of ideas that extensions of based the interest rates also really interesting nonstationarity and so there it going run on top of BCI and interested in taking a BCI device and or connecting 16 channels to lower forward and see if we can use it to classify individual finger movements is right on the system regret where a single view with that can make a hand flex in like an analog motion religious and see if we we can identify individual fingers so it just my idea of personal idea I was reading about the complex adaptive systems book and it seems like it is it used as a genetic algorithms in order to have to solve some problems in in nature but I want to invent of bringing to it so that like you go into the that adaptive system to make that there are some papers that actually combine 2 but was only 1 thing found on so I'd like to be able to do do that with the mathematical texts Mathematics can actually use used on I don't recall if you can't do that the religious view in order to keep things much faster and easier to code so I was trying to actually implement you know that the algorithms in the mathematical theory don't think that would be I would say about it has the most implementations this is due to a lot of interest in getting ready to grow and you you see in other words if you like of integers right you actually true so all the portable title and you try to provide always and already in the API tool that they always and all of all texts they all of this strings and if you go over the effects they have read to try all of the quality of life for years for free and so on and you want to use requires a I wouldn't really call on top of these guys up here in time and John working on the project 1 is the acoustic project that by listening to the noise of the false itself want you detect which falls in where it is and the other 1 the 1 you on something on the same dataset using this switch matrix on the beach on the EEG device instead of hard-coding them that you want you want to do the weight learning as part of its which matrix on the user has unexplored layer of and the 1st technology audio analysis that the audience the are others here Richard is interested in audio waveform analysis of certain things and I don't experiments with 2 and largely unsuccessful that's going on the radio so far I have to find work right this is this is all part of patterns and the idea is to build something is that there are patterns of in that we have learned that the concentrated on the status of the new jobs so and we include the data in such a way that you can see patterns and just predict you often that's not prices just direction the we did not nominally successful start all introduction so another possible the things you know and interested in analyzing content and inclusive protest so the model that includes the occasional of this is the probability of a lack of it's it's kind of of explaining the on the set of indices so you you can you know and hence that it might be wise burstiness especially in the back into that idea you want to think you create data running through we have tried to find the omentum employers because of that and it's very easy to make interactive assumptions about how the Danish these weights and what we call the and and so you have even in any possible if this is really useful more people to go up on the data link anything else so it's it's worth the gravity of the of the user you're circles for a long time interval and cellular it is wrong for OK so that was the problems to the great rate you often appears talk about the of of the access also with unit unit project and you make you all of which would have a I I I I did it
Demo <Programm>
Momentenproblem
Selbstrepräsentation
t-Test
Raum-Zeit
Computeranimation
Videokonferenz
Einheit <Mathematik>
Reverse Engineering
Trennschärfe <Statistik>
Gruppe <Mathematik>
Protokoll <Datenverarbeitungssystem>
Vorlesung/Konferenz
Hacker
Schnelltaste
Schwach besetzte Matrix
Biprodukt
Ereignishorizont
Rechenschieber
Scheduling
Quelle <Physik>
Verkettung <Informatik>
Registrierung <Bildverarbeitung>
Rechter Winkel
Projektive Ebene
Decodierung
Ordnung <Mathematik>
Aggregatzustand
Automatische Handlungsplanung
Physikalische Theorie
Bildschirmmaske
Digitale Photographie
Notebook-Computer
Stichprobenumfang
Zusammenhängender Graph
Logistische Verteilung
Informatik
Touchscreen
Videospiel
Open Source
Mailing-Liste
Vektorraum
Quick-Sort
Hill-Differentialgleichung
Integral
Objekt <Kategorie>
Mereologie
Hypermedia
Speicherabzug
Wort <Informatik>
Verkehrsinformation
Orientierung <Mathematik>
Gemeinsamer Speicher
Selbstrepräsentation
Zentraleinheit
Skalarfeld
Physikalische Theorie
Whiteboard
Code
Knotenmenge
Informationsmodellierung
Prognoseverfahren
Arithmetische Folge
Vorzeichen <Mathematik>
Notebook-Computer
Stichprobenumfang
Mustersprache
Vorlesung/Konferenz
Logistische Verteilung
Indexberechnung
Hacker
Softwareentwickler
E-Mail
Gerade
Fundamentalsatz der Algebra
Hardware
Protokoll <Datenverarbeitungssystem>
Computersicherheit
Gebäude <Mathematik>
Mailing-Liste
Physikalisches System
Binder <Informatik>
Wiki
Quick-Sort
Ereignishorizont
Rechenschieber
Energiedichte
System F
Menge
Flächeninhalt
Offene Menge
Chatten <Kommunikation>
Hypermedia
Mereologie
Wort <Informatik>
Projektive Ebene
Information
Wärmeleitfähigkeit
Addition
Punkt
Wasserdampftafel
Temporale Logik
Implementierung
Zahlenbereich
Kombinatorische Gruppentheorie
Technische Optik
Physikalische Theorie
Eins
Variable
Einheit <Mathematik>
Algorithmus
Prozess <Informatik>
Konditionszahl
Festspeicher
Mustersprache
Vorlesung/Konferenz
Optimierung
Schlüsselverwaltung
Fitnessfunktion
Resultante
Sensitivitätsanalyse
Prozess <Physik>
Natürliche Zahl
Selbstrepräsentation
t-Test
Bridge <Kommunikationstechnik>
Kartesische Koordinaten
Skalarfeld
Raum-Zeit
Internetworking
Übergang
Videokonferenz
Formale Semantik
Algorithmus
Regulärer Graph
Gruppe <Mathematik>
Mustersprache
Vorlesung/Konferenz
Gerade
Zentrische Streckung
Schwellwertverfahren
Datennetz
Computersicherheit
Gebäude <Mathematik>
Mobiles Internet
Stellenring
Quellcode
Bitrate
Mustererkennung
Frequenz
Ereignishorizont
Dialekt
Dienst <Informatik>
Rechter Winkel
Konditionszahl
Nuklearer Raum
Dateiformat
Projektive Ebene
Information
Decodierung
Ordnung <Mathematik>
Programmierumgebung
Folge <Mathematik>
Lochstreifen
Gruppenoperation
Besprechung/Interview
Interaktives Fernsehen
Trajektorie <Mathematik>
Physikalische Theorie
Data Mining
Informationsmodellierung
Netzbetriebssystem
Datentyp
Passwort
Zusammenhängender Graph
Inklusion <Mathematik>
Datenstruktur
Analysis
Relativitätstheorie
Validität
Hasard <Digitaltechnik>
Schlussregel
Quick-Sort
Programmfehler
Numerische Taxonomie
Druckertreiber
Mereologie
Wort <Informatik>
Matrizenrechnung
Gravitation
Punkt
Wellenpaket
Gewicht <Mathematik>
Freeware
Natürliche Zahl
Besprechung/Interview
Versionsverwaltung
Geräusch
Implementierung
Komplex <Algebra>
Physikalische Theorie
Richtung
Videokonferenz
Informationsmodellierung
Einheit <Mathematik>
Algorithmus
Zeitreihenanalyse
Prozess <Informatik>
Mustersprache
Randomisierung
Inhalt <Mathematik>
Indexberechnung
Inklusion <Mathematik>
Maßerweiterung
Grundraum
Analogieschluss
Analysis
Soundverarbeitung
Videospiel
Kreisfläche
Sichtenkonzept
Mathematik
Wellenform
Physikalisches System
Bitrate
Binder <Informatik>
Natürliche Sprache
Menge
Ganze Zahl
Anpassung <Mathematik>
Mereologie
Garbentheorie
Projektive Ebene
Wort <Informatik>
Ordnung <Mathematik>
Instantiierung

Metadaten

Formale Metadaten

Titel Kickoff (2015 Spring NuPIC Hackathon)
Serientitel 2015 Spring NuPIC Hackathon
Anzahl der Teile 19
Autor Taylor, Matthew
Lizenz CC-Namensnennung 3.0 Unported:
Sie dürfen das Werk bzw. den Inhalt zu jedem legalen Zweck nutzen, verändern und in unveränderter oder veränderter Form vervielfältigen, verbreiten und öffentlich zugänglich machen, sofern Sie den Namen des Autors/Rechteinhabers in der von ihm festgelegten Weise nennen.
DOI 10.5446/18053
Herausgeber Numenta Platform for Intelligent Computing (NuPIC)
Erscheinungsjahr 2015
Sprache Englisch

Inhaltliche Metadaten

Fachgebiet Informatik
Abstract Protocol and kickoff by Matt Taylor and Jeff Hawkins.

Ähnliche Filme

Loading...
Feedback