Archives for

Meet us at #WSAI17

The AI brains are coming to Amsterdam… the first ever “World Summit AI” taking place in Amsterdam. A great place for new inspiration and new connections.

Our team will be there for the full two days! Make sure to find Michiel (CTO) at our booth (right next to the main stage, can’t miss it!) or Jim (CEO) in the panel session on the second day.

Update
Also, check out some exciting news that came out of the event.

> Read full Meet us at #WSAI17 post


DARPA wil nieuwe chipsets voor AI

DARPA (Defense Advanced Research Projects Agency) heeft twee projecten aangekondigd voor de ontwikkeling van nieuwe computerchips die betere AI technologie mogelijk maken. De ontwikkeling van betere processoren dreigt namelijk vast te lopen, iets waar vooral de pioniers op AI gebied van beginnen te zweten.

Moore’s law

De afgelopen 50 jaar hebben kloksnelheden van processoren een enorme vlucht gemaakt. We staan er niet dagelijks bij stil, maar de gemiddelde smartphone is krachtiger dan supercomputers een paar decennia geleden. Gordon Moore, een van de oprichters van processorfabrikant Intel, zag al snel een regelmatig patroon in het tempo van de ontwikkelingen. Hij zag dat de complexiteit van computerchips, en daarmee de snelheid, ruwweg elke twee jaar verdubbeld. Dit gegeven kennen we tegenwoordig als Moore’s law.

Moore’s wet heeft het verbazingwekkend lang volgehouden. 50 jaar lang is er eigenlijk geen onderbreking geweest en pasten er meer en meer transistoren op een chip. De laatste jaren lijkt er echter een kentering te zijn. Net nu de vraag naar supersnelle chipsets explodeert door toepassingen als AI claimen experts dat het tempo stagneert. Brian Krzanich, de huidige Intel CEO, zegt hierover: “De laatste twee technologie transities tonen dat de cadans nu eerder twee en een half jaar is dan twee.” Dit komt mede doordat het niet goed meer mogelijk is om chips nog verder te verkleinen. Beneden de 4 nm doet het zogenaamde quantum tunneling effect zich voor. Dit wil zeggen dat electronen, op nano schaal, niet altijd op hun plek kunnen worden gehouden.

Het ERI programma

Om de ontwikkeling van AI, en een Amerikaanse AI voorsprong te behouden, zal deze horde toch genomen moeten worden. Om de ontwikkeling van processoren de benodigde boost te geven, heeft DARPA het ERI programma in leven geroepen.

Het ERI programma is verdeeld in drie onderdelen, zogenaamde thrusts. Elk van deze thrusts zal er op gericht zijn om specifieke vraagstukken aan te pakken. hierbij valt te denken aan het gebruik van onconventionele materialen op chipsets, nieuwe ontwerp technieken of zelfs een geheel nieuwe architectuur. Het zou dus zomaar kunnen dat de processors van de toekomst van de huidige Neumann architectuur zullen zijn afgestapt.

The ERI will kickstart new chips for better AI

De nieuwe space race

Het lijkt er op dat het vergroten van de AI voorsprong, met het ERI programma, weer hoger op het prioriteitenlijstje is gekomen. Zeker nu China grote investeringen doet en met rappe schreden nadert. De strijd om de beste AI algorithmen wordt dan ook weleens gekscherend “de nieuwe space race” genoemd. Het al me dan ook niet verbazen als Europa zich binnenkort ook op het toneel meldt. Belangrijker dan de strijd tussen landen is misschien wel wat voor moois bedrijven en NGO’s met vergaande AI mogelijkheden kunnen doen. Wat dat betreft houd de analogie met de space race aan. Ook toen werden er veel technologieën ontwikkeld die uiteindelijk bij de consument terecht zijn gekomen.Die muis die je vast hebt, die laptop waar je op werkt, zijn slechts twee voorbeelden van technologieën die bij NASA vandaan komen. Wij kijken in ieder geval uit naar alle nieuwe mogelijkheden die verbeterde chipsets zullen bieden.

Bron: DARPA

> Read full DARPA wil nieuwe chipsets voor AI post


Fraiday 6: AI for Good

Fraiday is the monthly meet-up for professionals who are AI-curious. We have a simple format: involving beers, a theme, lively debate and a group picture.

Fraiday #06 AI for Good 29-09-2017 photo by Mike Breeuwer 19

On 29 September we gathered at the Startup Village to hear from an all-star at the 6th edition of Fraiday. YouTube celebrity Siraj Raval, Jan Kees Schakel from Sensing Clues and Rutger Hofste from the World Resources Institute took the stage to shine some light on the ways that NGO’s are using Machine Learning to improve the world.

The stacked containers of the Startup Village at the Amsterdam Science Park provided the decor for this Fraiday. As people gathered at the bar in anticipation of the speakers it quickly became clear that Fraiday’s turnout is steadily increasing. 85 AI enthusiasts came to hear how they could make a difference.

Aigency’s CEO Jim Stolze started the evening by introducing the AI for Good initiative. With the help of the SIDN Fund, Aigency started a matchmaking initiative between NGO’s and AI talent. This will enable NGO’s to harness their data and leverage the insights that can be uncovered from this data.

Siraj Raval at Fraiday #06 AI for Good photo by Mike Breeuwer8First up as speaker was Siraj, he started his talk with the ways that you can self educate as AI curious person. Perhaps counter-intuitively, Google should not be your starting point. That role is left for Github. Later he harped on the grand potential that the combination of Blockchain and AI can have for the world. “The infrastructure we’re building right now. Is gonna have a millionfold impact.” In this scenario, cryptocurrencies allow for everyone to get paid directly for the data they provide. Blockchain also prevents one person from gaining absolute power based on the data set that he or she controls. A very important point, because as Siraj said: “Absolute power is still absolute power. Regardless of the wielder of that power.”

Jan Kees Schakel at Fraiday #06 AI for Good 29-09-2017 photo by Mike Breeuwer17Jan Kees Schakel from Sensing Clues introduced an example of AI helping the world. In this case that meant the results of a wildlife hackathon, where AI has been used to protect wildlife from poachers. It turns out that humans are pretty noisy creatures due to the tools that they bring. Sound Intelligence Systems made up an early warning network that gave authorities intelligence of the moves that poachers were making.

Is there any better way to start of your talk than giving away jars of Nutella? It is no surprise that Rutger Hofste made some Fraiday-friends during his intro by doing so. Still, he had quite a serious topic to discuss. Palm oil, one of the ingredients of Nutella, is in fact damaging to the environment due to the enormous plantations that are necessary to produce it. Luckily, now that satellite imagery has been made available by NASA and ESA, AI developers can help map the extent of the damage done and hold those responsible accountable.

After all of this educational content it was time for some drinks and snacks. The Coffee Virus provided some delicious rolls, and someone ordered a couple of pizza’s. It was really cool to see that conversations went on until after closing time. The pictures of the day are on Aigency’s Facebook page. Make sure to take a look!

Next Fraiday, we’ll be exloring the ways that AI can be used in a business context. Want to stay up date? Make sure to Follow, like and subscribe on Twitter, Facebook and Meetup.

> Read full Fraiday 6: AI for Good post


Adoptie van robotica werkt alleen als alle stakeholders aan boord zijn

Gezamenlijk plannen en dialoog tussen diverse beroepsgroepen en stakeholders is essentieel bij het implementeren van AI en robotica. Onderzoek van het VTT Technical Research Centre of Finland, een Fins publiek instituut, in de zorgsector heeft dit uitgewezen.

Vanwege de vergrijzing en het personeeltekort in de zorg kijken veel zorgaanbieders naar robotisering om het serviceniveau op peil te houden. Gezien de aard van het werk is robots inzetten ter ondersteuning echter niet zo straight-forward als dat voor sommige andere projecten is. Zorg en welzijnswerk is in beginsel mensenwerk.

De onderzoekers hebben de implementatie van een state of the art gerobotiseerd logistiek systeem getest. Net als in Nederland, staat de implementatie van geautomatiseerde logistieke systemen nog in de kinderschoenen. Het Seinäjoki Hospitaal zal met 5 tot 8 bezorgrobots dan ook één van de pioniers zijn. Tijdens de studie is een pilot gedraaid met 2 van deze robots.

Ziekenhuizen hebben normaliter 24/7 te maken met leveringen, het systeem heeft de beschikbaarheid van voorraden verbeterd en congestie op de gangpaden aanzienlijk verminderd. Ook zijn de personeelskosten en de fysieke belasting van het werk gedaald. Het personeel is ook overwegend positief over de robots.

Natuurlijk heeft de introductie van een dergelijk systeem verschillende effecten op de beroepsgroepen die in een dergelijke organisatie samenkomen. Vooral de mate waarin men het gevoel heeft controle over het eigen werk te hebben, is samen met waardering voor het werk een belangrijke factor.

Uit de studie is gekomen dat zorgvuldigheid bij het plannen, de pilots en de implementatie essentieel zijn. “Terwijl het systeem wordt uitgebreid met nieuwe robots en typen leveringen, zal er nog meer begeleiding, communicatie en dialoog nodig zijn dan voorheen. gezamelijk plannen met de verschillende stakeholders zorgt er voor dat de implementatie zo soepel mogelijk verloopt. Zo is het makkelijker de beoogde voordelen te realiseren.” zegt Inka Lappalainen, Senior Scientist van het ROSE project hierover.

Bron: ROSE project

> Read full Adoptie van robotica werkt alleen als alle stakeholders aan boord zijn post


Fraiday 5: Let’s Get Practical!

Fraiday is the monthly meet-up for professionals who are AI-curious. We have a simple format: involving beers, a theme, lively debate and a group picture.

Fraiday #05 Lets get Practical 25-09-2017 40

Photos courtesy of Mike Breeuwer. Click the photo for the whole album!

The 5th edition of Fraiday was all about practical application of AI. Bart Vredebrecht from Aiir Innovations and our very own Michiel Berger were there to educate participants on all the ins and outs of AI projects.

For a change of scenery, this edition of Fraiday was at Freedomlab. Upon entering it was immediately clear that it is a place that defies definition. Co working-space, maker-space and innovation hub are all lables that sort of fit the bill at this “Institute for Redefinition” that is all about improving the way we life, work and relate to one another.

Bart gave us an overview of the triumphs and pitfalls that a real life AI project entails. He is the CEO of Aiir Innovations, an AI company that helps airlines with inspection of aircraft engines. He highlighted that a team of human + AI will outperform either of them alone.

After this introductory talk the group split into two breakout sessions. One group, meant for people who are already implementing AI, went with Michiel and discussed practical issues relating to that. The other group, which consisted of the AI curious, was introduced to the Machine Learning Canvas by Bart. The Machine Learning Canvas, which is based on the Business Model Canvas,

Fraiday #05 Lets get Practical 25-09-2017 47

Photos courtesy of Mike Breeuwer. Click the photo for the whole album!

After all that learning the group needed some nourishment, which duly arrived. The post-event discussions were held under the enjoyment of pizza and drinks. If you’d like to see the pics that were made during the day. Click here for the Facebook album.

At the next Fraiday, on 29 September, we’ll find out about all the different ways that AI can be used for good in non-commercial settings. Curious about the possibilities? Join us!

> Read full Fraiday 5: Let’s Get Practical! post


#FRAIDAY 29 SEPT: AI FOR GOOD

Are you interested in Artificial Intelligence and creating a better world? Come to the next Fraiday on the 29th of September!

How can we use AI for the greater good?

During the next Fraiday we’ll discuss how Artificial Intelligence can be used for humanitarian causes. Expect many students, startups and NGO’s + a surprise speaker. Warning: you might be accused of being overly “optimistic” after you attend this meetup.

  • Date: 29th Sept 2017
  • Time: 6 pm
  • Location: Venture Studio, Science Park, Amsterdam

Interested? Sign up here!

 

> Read full #FRAIDAY 29 SEPT: AI FOR GOOD post


Zeynep Tufekci

The TED talk of the month discusses big questions about our societies and our lives, as both algorithms and digital connectivity spread. Complex algorithms that watch, judge and nudge us.

About the speaker
Zeynep Tufekci is a contributing opinion writer at the New York Times, an associate professor at the School of Information and Library Science at University of North Carolina, Chapel Hill. Her book, Twitter and Tear Gas: The Power and Fragility of Networked Protest, was published in 2017 by Yale University Press.

> Read full Zeynep Tufekci post


Live from Taiwan!

The largest computer fair in Asia: Computex. Not surprisingly AI was an important theme this year! Our COO Eiso Vaandrager represented Aigency in this browsing environment. He occupied a popular booth in the AI-section and took the stage on day #2.

Computex: AI“We spoke with a lot of different interesting parties and presented the AI concept of Aigency. This resulted in conversations with a lot of interesting Taiwanese companies where cooperation is possible. All the big players where there, like IBM Watson, Google, Intel, and NVIDIA. They were all there to see what is happening in the AI area” says Eiso.

Taiwan is a huge producer of electronica and leader in the AI area, therefore it was an interesting place to present Aigency. Another Dutch company called Travis the Translator, was also present. Travis is a bot that speaks 80 different languages.

Moreover, Eiso says he was “definitely impressed by the whole event, and once again overwhelmed how fast all this is going. Even if you are following the AI updates, you are still behind”.

An interesting insight was that the gap between some big companies who are really going for it, and some others who are staying behind is getting bigger and bigger. An opportunity for us is that we can help these companies, because they have no clue where this is going and they will wake up in a new world.

Besides that, it is remarkable that in Asia there is much more willingness to go and experiment and implement with AI versus here in Europe.

In Europe, there is so much discussion. We have to be careful with this, because in Asia they understand the sense of urgency better. Here in Europe we also have to experiment, that is the way you learn and be able to implement things that are working. In the Netherlands when you have a conversation about AI, it is most of the time a shallow conversation about: “we have data”. In Asia however the conversations are far more in depth, about data and privacy for example. They are one step ahead of us.

A last quote from Eiso: “It was a really nice experience in Taiwan, especially since they are totally electronica crazy over there!”

 

> Read full Live from Taiwan! post


Maurice Conti

The TED talk of the month comes from designer, futurist and innovator Maurice Conti. At TEDxPortland he showed how robots and humans will work side-by-side to accomplish things neither could do alone.

About this speaker
Maurice Conti is the Director of Applied Research & Innovation at Autodesk. He also leads Autodesk’s Applied Research Lab, which he built from the ground up. His team’s research focuses on advanced robotics, applied machine learning, the Internet of Things and climate change/sea level rise.

> Read full Maurice Conti post


The perfect selfie, assisted by AI

Everyone has experienced the feeling of seeing yourself appear on the screen of their phone unexpectedly, because they opened the camera-app with the selfie camera enabled. Not pretty. Well, today I learned that this is due to an effect called perspective distortion.

According to wikipedia, when shooting a portrait foto and fitting the same area inside the frame:

The wide-angle will be used from closer, making the nose larger compared to the rest of the photo, and the telephoto will be used from farther, making the nose smaller compared to the rest of the photo.

Photographers have known this for ages. That’s why professional portraits are usually shot from a distance, using a telephoto lens to fit the subject’s face in the frame. But us civilians mostly capture their faces with a selfie camera, which uses a wide angle lens. Maybe if we had 4m long selfie sticks, we could do something about it. But that doesn’t seem very practical to me.

The technology

Researchers from Princeton and Adobe have developed an algorithm that can adjust camera distance in post production [1]. They achieve this by estimating a 3D model of the face, in which camera position and orientation are included, and fitting this to the 2D image. If you then manually change one variable in this model, the algorithm calculates the expected changes in the remaining properties and the result is projected onto a 2D image corresponding with the new camera position.

The representation of the face is obtained by automatically location 66 facial features around the eyes, nose and chin. For this step, the researches employ existing technology by Saragih et al (2009) [2]. Because the detector they use doesn’t find key points on the ears and top of the head, these points have to be added manually. Those points are necessary to incorporate the ears and hair into the model. Without those, warping would result in an uncanny result where the perspective of the face changes, but that of the hair and ears stays the same. Read More

> Read full The perfect selfie, assisted by AI post