Synaptech continues the journey of building the newest AI community in Europe, one step at a time. Wishing to better understand the real applicability of Artificial Intelligence, we`ve gathered around our event brilliant minds from all over the world. We like to think that Synaptech will be a stepping stone, at least a small one, into the evolution of AI in Europe.

20 speakers join us for Synaptech

At the end of April, we announced the first Synaptech speakers, as a short glimpse into our upcoming event. All of them and many more, from passionate researchers, to innovative CEO`s and university professors will help to decipher and better understand the usage of Artificial Intelligence.

More and more companies benefit from AI

Did you know that between 34% and 44% of global companies are using AI in their IT departments to resolve employees’ tech support problems or enhance production? A survey on Harvard Business Review shows that by 2020 AI will be a crucial part in every company. Basically, AI will have the biggest impact on the back-office functions of IT and finance & accounting in over 50% of the companies that have already embedded AI into their business.

According to Atomico’s 2016 report on the state of technology, the European tech industry is increasingly focused on deep tech. If in September 2015 there were just 146 articles reporting on AI, just one year later the number rose to 1.112 mentions (almost 8 times higher). This shows an increased interest in the field of AI all around Europe.

Learn from leaders, researchers and innovators in the AI field

As AI is expanding in an accelerated way, so we intend to do with our event. We invited leaders, professors and dedicated innovators to shed a light on the implications of AI in the real world. We are enthusiastic to announce:

Sebastian Wieczorek, Head of SAP Machine Learning Foundation. Sebastian is leading a team of data scientists and developers in Berlin, Walldorf and Singapore to build SAP’s machine learning platform for delivering Application Intelligence.

Cyrus Moazami-Vahid, Principal Deep Learning Solutions Architect at Amazon. In the past 20 years, Cyrus has delivered innovative solutions with track-record of success stories mostly in entrepreneurial environment. In The past 3 years he has dedicated his work to generating Big Data initiatives and solutions.

Manuel Koelman — Co-Founder of PIRATE.global. In 2010, he launched “The Pirate Summit”, Europe’s largest invitation-only conference for early-stage startups, investors and corporate executives.

Alexa Gorman, Global VP SAP.io Fund at SAP. With years of experience at SAP, Alexa has great experience in the fields of Corporate Strategy and Business Development

Arnaud Muller, Founder & CEO of Saagie. Muller founded Saagie, an end-to-end data platform that boosts your business with artificial intelligence in 2013.

You can find out more about the speakers joining Synaptech on our website.

Engage in discussions about AI

On the first day of Synaptech we’ve prepared four workshops. The first one, STM Networks in MXNet for sentiment analysis will be held by Cyrus Moazami-Vahid, Principal Deep Learning Solutions Architect at Amazon. This talk gives a general description of a network that provides sentiment analysis for IMDB movie reviews, network architecture, and implementation in Apache MXNet.

Discover startups that innovate by putting AI at work

Our goal is to discover and comprehend the newest AI developments. Thus, we’ve also developed a startup competition, in partnership with Axel Springer Plug and Play, dedicated to AI-focused startups. Our call for subscriptions attracted 50 applicants from countries all over Europe: UK, Netherlands, Belarus, Poland, France, Norway, Israel and of course, Germany. Keep an eye on our Social Media channels to find out which startups will pitch on the Synaptech stage!

Don’t miss a moment of Synaptech and book your seat now!

With just a little over a month until the event, this is the perfect moment to reserve your ticket. You can choose from three types of tickets: Basic — 199.00 €, Premium — 499 € and VIP — 849 €.

The Basic ticket offers access to 2 out of 4 Machine Learning Workshops on September 21st. For a full Synaptech experience, try the Premium and VIP options. Both assure access to the Machine Learning Workshops, AI Startups Competition and Conference. This type of tickets will be available until 12.09.2017. Afterwards, the prices will rise for the “Last Minute” tickets: Basic — 249.00 €, Premium — 599 € and VIP — 999 €.

Find out more about Synaptech taking place between September 21 and 22, at Kühlhaus Berlin by following us on Facebook and Twitter. For the latest news on speakers and tickets sign-up for the newsletter.

August brings exciting news for Synaptech | Innovators and leaders of the AI field are joining the… was originally published in Synaptech on Medium, where people are continuing the conversation by highlighting and responding to this story.

The term “Machine Learning” has become so common in the field of AI, that most of the time people mistake it as a synonym for AI. But if we were to imagine AI as a vehicle, we could say that Machine Learning is more or less its engine.

Machine Learning is a type of AI which enables software applications to become better at predicting outcomes, without programming them explicitly to do so. To receive the desired results, you have to build algorithms that can receive data and use statistical analysis to offer their output within a requested range of probabilities. For instance, if you want a program to tell you 10 dystopic scenarios based on 100 movies or books, it will do so.

There are two types of algorithms present in Machine Learning:

• Supervised: they require external input and a desire output, in order to polish the accuracy of predictions. After the algorithm’s training is complete, it will continue to apply what it learned to the new data.
• Unsupervised: they don’t need any training and use an iterative approach — deep learning, to analyze data and to submit conclusions. These types of algorithms are used for complex processing tasks.

How it works

Machine Learning is focused on Data, like any AI sub-type. Thus, ML searches through data and looks for patterns and adjusts its activity accordingly. To be more exact, do you know how when you are watching some new cool series, Netflix offers you another movie or series suggestion? Or when you are shopping, you can find new recommendations? Oh well, blame it on ML!

All of this is happening because recommendation engines use ML to personalize their ad delivery in real time. Apart from convincing you to buy new interesting stuff, Machine Learning is used for fraud detection, to filter spam, to identify any network security threat and also to build the newsfeed we all know and love (or not) on Facebook.

This is what Machine Learning means, on short. There are plenty of other things we could mention about it. If we missed something, please drop us a line on Twitter, Facebook or in a comment. And if you are passionate and you want to learn more about AI, register for Synaptech’s hands-on Machine Learning workshops and conference where thriving thought leaders will share their experience with AI! More details you can find here.

See you next time in our newest article about Natural language processing!

Understanding algorithms: a short introduction to Machine Learning was originally published in Synaptech on Medium, where people are continuing the conversation by highlighting and responding to this story.

What if you could become an artist in less time than it would require, let’s say approximately 3 years? All you had to do is to make a simple sketch, press some buttons and then poof, you have created a piece of artwork. Does this sound Impossible? Not anymore.

Thanks to the great evolution of Computer Vision and all other cool algorithms that were developed afterward, AI is now slowly, but surely, learning how to design images, based on input. The process evolves with heavy steps, since the major problem with computers is the way they understand images, and also because human appreciation to what looks good is subjective and personal.

Now, let’s see a few of the algorithms that can help you turn your black and white sketch into a full image:

Sketch-based image retrieval

This can be achieved by a transformation of GoogLeNet architecture into a triplet network. Thus, it allows the network learn across both the sketch and the image domain, and also a shared feature space between the two.

Image synthesis with Neural Networks

Using deep Convolutional Neural Networks (dCNN) for more productive tasks, such as texture synthesis. They could be used with a VGG-19 network, to transform the texture features of the “style” image within each layer of the network into a set of Gram metrics whilst capturing high-level of the content image.

Another option is to use Generative Adversarial Networks. As we presented them in the past article, there are two networks, one that plays the role of creation and the other of the discriminator. Thus, if trained enough, they could be able to generate more accurate image than the dCNN one.

MRF-based image synthesis

This type of method implies the use of generative Markov random field models, where dCNNs would work both for photorealistic and non-photorealistic image synthesis. Instead of using Gram matrices, the MRF maintains local patterns of the specified style, while using the same VGG-19 network, but trained only on images.

So, as we can see, image synthesis still has a long way to go. Or not, varying on each one’s opinion. But AI makes great progresses and, maybe sooner than we expect, we could see a remake of Star Wars with puppies instead of Jedis. The entertainment industry will certainly have a bright future.

We will continue our series of cool AI-terms article next week with “Machine Learning”. Until then, drop us a line on Synaptech’s Facebook or Twitter. See you!

From sketches to photo realistic images: an introduction to image synthesis was originally published in Synaptech on Medium, where people are continuing the conversation by highlighting and responding to this story.

Imagine that you are a painter. Not any kind of painter, but one that is specialized into, how to put it gently, not creating new things, but forging things. Let’s call you G. Now, imagine opposed to you, there is this guy, D, who is an art critic — a very talented one, that can spot a fake painting from miles away. Now, let’s imagine you want to show him some of your “Monet’s”.

At the beginning, G’s whole purpose is to create fake Monets. Sometimes, D falls for them, sometimes he doesn’t. But, as time passes by and D starts to see more and more original examples, he becomes better at detecting fakes. Since G starts having a harder time to, let’s say, fool D, he has to become better. Thus, he slowly starts to provide better forges. On short, this is the idea behind Generative Adversarial Networks or GANs.

What are Generative Adversarial Networks

GANs is a new type of generative model, which are a branch of unsupervised learning techniques in Machine Learning.

GANs contains two networks that live in a constant conflict — thus the adversarial term, a generator (G) and a a discriminator (D). As in anything related to Machine Learning, GANs can be trained with examples, such as images, and there is an underlying distribution (x) that governs them. Thus, G will generate outputs — or create new stuff, and D will decide if they come from the same distribution of the training set or they are, you guessed, fake.

How do GANs work — technically

Generative Adversarial Networks are like this: G, will start from some noise (z), and the images it generates are G(z). D, on the other hand takes from the real images (x) and the fake ones, from G, and classifies them as D(x) and D(G(z)).

What’s interesting about them is that both are learning at the same time. Once you train G with enough input, it will know enough about the distribution and it will be able to generate new samples that share very similar properties. And as you train D, it will sense if the objects from the image are real or not.

Some real applications of GNAs

• Image generation from input samples
• High resolution image generation from a lower one
• Interactive image generation — iGANs
• Diagrammatic Abstract Reasoning
• Image super resolution
• Image in painting
• Semantic segmentation
• Video generation
• Text to image generation

Until the release of our next article, which will be about image synthesis, you can check our Facebook and Twitter accounts, where we will post other great things about AI. In case we have missed some important piece of info about GANs, do not hesitate to leave us a comment.

Counterfeits, beware! These new neural networks will catch you was originally published in Synaptech on Medium, where people are continuing the conversation by highlighting and responding to this story.

Imagine that, at some point in life, scientists gather together to create the ultimate weapon to save humanity: an artificial brain. A product so complex that it can mimic everything a human brain can do, but it’s artificial instead of organic.

Now, dream no more! This whole concept is one step away from becoming real, through Deep Learning. Maybe more than a step away, more like a few. But, close, nonetheless.

What is Deep Learning?

Deep learning is also known as deep structured learning or hierarchical learning, and it uses artificial neural networks, or ANNs for short. ANNs are basically nothing else but a pair of hidden layers, with a design inspired by the neurons in the central nervous system of our brain. For short, ANNs are nothing else but artificial neurons.

Artificial Neural Networks are presented as a system of interconnected points, with the purpose of exchanging messages between each other. These connections carry numeric values, which can be tuned based on experience, thus becoming capable of learning.

For instance, as a biological neuron has dendrites, axons and synapses, an artificial one is composed of nodes, inputs, outputs and weights. For instance, like in this photo.

How do ANNs work?

You can stack layers of neurons on top of each other. For example, the lowest layer takes the raw data — the background image. The next layer will find more details, since it can learn more data from the previous layer. And so on, until you re-create another image.

How will this help me in the future?

Let’s say you’re a graphic designer and you really need to know all the hex codes presented in an image. Since your time is short and the idea of colour-picking every single pixel in the image is a hassle, you can design an ANN to pick them for you. All you have to do is to feed it with the raw image.

In the near future, ANNs will be able to pick different kind of datasets, such as raw text and numbers. If you are interested in learning how to develop Machine Learning algorithms, ANNs will prove to be useful in creating them.

This all about Deep Learning, for the moment. But our article series doesn’t end here. Stay tuned for our next article about Generative Adversarial Networks. Meanwhile, if you are tempted to learn more about AI from thriving experts, reserve your ticket to the newest AI-conference based in Berlin, Synaptech!

Will computers be smarter than us? Possible, through deep learning was originally published in Synaptech on Medium, where people are continuing the conversation by highlighting and responding to this story.

What is a Decision Tree?

A decision tree is a predictive algorithmic model that goes from the observations of an object (branches) to the conclusions about the object’s value (leaves). Its purpose is to predict modelling approaches in statistics, data mining and machine learning.

Regarding the types the decision tree is used, the way it is implemented can vary. For instance, for a decision analysis, the decision tree can be chose to represent in a visual and explicit manner the decisions, while in data mining, a decision tree describes data. In the resulting classification, the tree can be justified as an input for decision making.

Types of Decision Trees

Just like any other ML algorithm, the decision tree has plenty of types. There are two main types managed in data mining:

• Classification tree — its goal is to create rules that can be explained and translated into software language. It records, labels and assigns variables to classes. A classification tree is built through a process also known as binary recursive partitioning. The process implies splitting data into partitions and then splitting them up further on each of the branches.
• Regression Trees — this type deals with a continuous goal variable. The methods perform induction by means of an efficient recursive partitioning algorithm. Briefly, regression trees provide good predictive accuracy on many domains. However, the simple models used in their leaves have some limitations regarding the kind of functions they are able to perform.
• Decision stream — it’s a diagram which illustrates the information or process flow in a system or subsystem.

If you want to add some details or share your opinion this algorithm, feel free to share them with us in a comment. Meanwhile, stay tuned for our next article about AI terms, Generative adversarial networks.

Decision Tree — the smartest tree we know was originally published in Synaptech on Medium, where people are continuing the conversation by highlighting and responding to this story.

AI enthusiasts, we have great news for you! Since Synaptech is an AI-focused event, addressed to the startup community, we added a new layer to help promoting the cool tech ecosystem from the region: an AI competition for startups who redefine innovation, in partnership with Axel Springer, Plug and Play.

If you have an AI-focused tech startup, we want to let you know that the application form is now live on F6S and the deadline for registration is August 16. The proper competition will take place in September 21–22, at Kühlhaus, in Berlin.

You will have 4 minutes to pitch your product and 2 minutes for live demo, in front of investors, experts and VIP guests.

Applications criteria:

• Products focused on Artificial Intelligence;
• Startups need to have an MVP;
• Currently fundraising;
• Functional live demo.

Both applications and pitching need to be in English.

Be bold, be brave and apply!

More details will be added on our website, later on. For any question, feel free to e-mail us at contact@synaptech.ai or DM us on Twitter or Facebook.

Do you have an AI product? Apply to Synaptech’s Startup Competition! was originally published in Synaptech on Medium, where people are continuing the conversation by highlighting and responding to this story.

What are Convolutional Neural Networks?

Convolutional Neural Networks (or CNN) are mainly used for Image Classification and are the core of Computer Vision. With their help, technology has made major breakthroughs, such as allowing self-driving cars to analyze and predict pedestrians’ movements. CNNs are also used for simpler tasks, such as Facebook’s auto-tagging.

Convolutional Neural Networks are comprised of convolutional layers, followed adjacently by connected layers — similar to a multilayer neural network. A CNN has the capability to understand 2D input, an image or a speech signal.

How do CNN work?

They are built with local connections, where each region of the input is connected to a neuron in the output. Because there are so many layers, each of them has a different filter and combine their result, or pooling (subsampling). What’s even cooler about Convolutional Neural Networks is that in the training phase, they automatically learn the values of their filters on the task you want them to perform.

For instance, if you program a CNN to detect something inside an image, it will learn to detect the corners from raw pixels in the first layer, then use them to detect simple shapes in the second layer. Afterwards, in higher layers they use shapes to understand fine features, such as a facial shape in higher layers. The last ones act as classifiers that use these features.

CNN has in composition two aspects worthy of your attention: Location Invariance and Compositionality. The easiest way to explain them is this: let’s say you want to classify whether there is a tiger or not in an image. Since you are sliding your filters all over the image, you don’t care where (location) in the image the tiger is positioned. Why? Because pooling does not change your composition when it translates, rotates or scales it, therefore it communicates the exact shape. Afterwards, each filter which helps composing the output, registers each pixel from the lower-level feature and transports it into the high-level representation (compositionality). Think about it as a very attentive Xerox that doesn’t miss anything when it copies your paper.

All in all, a CNN helps Computer Vision to be more intuitive and to build shapes from corners and complex objects from shapes.

How can they be used?

As we have seen, CNN are at the core of Facebook’s auto-tagging images. Of course, they can be used for more things, such as to analyze what beings and objects are in a photo, for educational, research or other purposes.

If you have this great idea of improving Machine Learning algorithms or if your startup does this already, book your seat for Synaptech this autumn. We have AI experts and a cool competition for startups, all in the same place. Also, you can stay tuned for our next article in the series: Decision Tree — a tree like no other.

Do computers silently judge your photos? was originally published in Synaptech on Medium, where people are continuing the conversation by highlighting and responding to this story.

Does the thought of being watched by machines make you quiver? Knowing that out there, hiding beneath a tree’s branch is a camera, recording every step you take, having knowledge of who you are and what you have done. It does seem like an adaptation of 1984 or other dystopian novels, but it might become reality through Computer Vision.

First of all, don’t worry and please don’t start throwing rocks at software developers for implementing it. Although this is the dark side of it, Computer Vision will be helpful to humanity in the upcoming years. Until then, let’s see what it is and what can it be used for.

What is Computer Vision?

It is the digitized version of the human vision. Computer vision is an algorithm that teaches a computer to reconstruct, interpret and understand a 3D scene from its 2D images as present structures in the scene. More exactly, how to tell the difference between a person, an animal and an object, through a camera lense, and it can be implemented in robots which are in the control of automatic lines and more.

Although the human vision is difficult to replicate, because it needs to replicate the visual cortex, the retina and the brain itself, it is somehow limited to the visible spectrum, while the one generated on computer can see further. But, the main problem is some images might be noisy and not offer enough information.

What and who are using it?

We know now what Computer vision is and on what basis its implementation can be made. But what companies do use it and on what exactly are they focusing the algorithm?

Eye tracking. What if you would have another eye in your own eye? Smart Eye offers that, a non-invasive eye, eyelid and head-tracking technology for a wide range of industries, but mostly for automotive, aviation and aerospace.

Object Recognition. One cool startup that developed a software that can recognize thousands of categories, tags, objects and images is Clarifai. The technology’s core is a deep learning API used in the development of a new generation of intelligent apps, including advertising, eCommerce and more.

People tracking. There it is, the shiver down the spine. Sighthound’s purpose is not to spy of humanity, but to be aware of the happening events. It can distinguish humans from objects and also has the ability of analyzing videos, in case of accidents, crimes, fights and more.

Medical imaging. Mirada Medical’s new development is changing the way diagnosis is offered through images. Their software is highly used in radiology, radiation oncology and other medical fields.

Face recognition. Goodbye, door keys! Thanks to Chui, the intelligent IoT doorbell that uses facial recognition to offer a secure, keyless door, now you don’t have to worry about getting locked outside.

Other domains where Computer Vision can be used are agriculture, AR, autonomous vehicles, biometrics, forensics, industrial quality inspection, gesture analysis, image restoration, pollution monitoring, remote sensing, process control and security and surveillance.

The next article regarding Artificial Intelligence glossary terms is convolutional neural networks. Until then, you can be kept up-to-date with Synaptech’s new speakers and other news, on our Twitter and Facebook account.

Dystopian or not, Computer Vision will help humanity was originally published in Synaptech on Medium, where people are continuing the conversation by highlighting and responding to this story.

This is where Synaptech steps in

Synaptech is focused on the real applicability of Artificial Intelligence. Its aim is to connect the most prolific-AI based startups with investors around the world, and corporations to the latest innovation, in 21–22 September.

A great event cannot take place without enlightened minds. Do you know what Jean-Francois Gagne, David Kelnar, Luming Wang, Friederike Schüür, Holger Weiss and Peter Yared have in common? Apart for being the first confirmed speakers for Synaptech, they are experts in the field of Artificial Intelligence, where they worked for years and nonetheless, they want to share their valuable experience with you.

Learn AI from Applied Researchers

Jean-Francois Gagné has co-founded with Yoshua Bengio, Element AI the world’s biggest applied research lab that develops AI-first solutions and technology to solve company challenges and boost productivity. Being a seasoned entrepreneur, he also founded and successfully exited two AI and operations research companies.

Friederike Schüür is the Director of Data Science at Fast Forward Labs, an applied machine intelligence and advising company in Brooklyn, NY. Their capabilities include diving into machine learning, building fully functioning prototypes exploring state-of-the-art technology and advising companies on how to get ready for the future (ML and AI).

Learn AI from Automotive Titans

Luming Wang is the Head of Deep Learning at Uber where he leads the development of the world’s leading Deep Learning platform, that supports Uber’s rapid growth.

Holger Weiss is the CEO of German AutoLabs and has 15 years track record in managing technology-driven companies with focus on consumer-centric services. Currently he is building an AI digital co-driver, in order to create a safer and smarter mobility experience.

Meet investors and CEOs passionate about AI

David Kelnar is the Investment Director & Head of Research at MMC Ventures, where he made an in-depth research on UK’s AI startups landscape. He has 8 years of entrepreneurial leadership experience in early stage companies, and is an advisor to a range of early stage ventures.

Peter Yared is the Founder & CTO at Sapho, the Google Now for the Enterprise. Previously, he was the CTO/CIO of CBS Interactive and the founder of 4 enterprise infrastructure companies. He is also the inventor of several patents on core Internet infrastructure including federated single sign on and dynamic data requests.

During Synaptech, apart from the conference, where the presented thought leaders will discuss about the latest developments and share their valuable insights, you will also be able to attend hands-on machine learning workshops and an international competition, featuring AI startups with disruptive products.

Embark in the world of AI and book your ticket for Synaptech now! For our first attendees we have a special offer: 2 tickets at the price of 1, available until April 30.

From applied researchers to CEOs passionate about AI: a glimpse of Synaptech’s first speakers was originally published in Synaptech on Medium, where people are continuing the conversation by highlighting and responding to this story.