Featured

Artificially Intelligent Painters: can deep learning AI create the next Mona Lisa?

Neural Style

If you have ever used Instagram or Snapchat, you are familiar with using filters that alter the brightness, saturation, contrast, and so on of your images. Neural style, a deep learning algorithm, goes beyond filters and allows you to transpose the style of one image, perhaps Van Gogh’s “Starry Night,” and apply that style onto any other image.  

Neural style, one of many models available on Somatic.io, uses a deep neural network in order to separate and recombine content and style of any two images. It is one of the first artificial neural networks (ANNs) to provide an algorithm for the creation of artistic imagery.

convolutional neural network

How Does it Work?

The model is given two input images, one that will be used for styling, the other for content. At each processing stage in the convolutional neural network’s (CNN) hierarchy, the images are broken into a set of filtered images. While the number of different filters increases along the processing hierarchy, the overall size of the filtered images is reduced, leading to a decrease in the total number of units per layer of the network.

The above figure visualizes the information at different processing stages in the CNN. The  content reconstructions from lower layers (a,b,c) are almost exact replicas of the original image. In the higher layers of the network however, the detailed pixel information is lost while the high-level structures and details remain the same (d,e). Meanwhile, the model captures the style of the other input image on top of the content CNN representations. Then, the style representation draws connections between the different features in different layers of the CNN. The model then reconstructs the style of the input image on top of the content representations within each of the CNN layers. This creates images that match the style on an increasing scale as you move through the network’s hierarchy.

convolutional neural network layers

Try It Out!

Experiment with the model for yourself. All you need to do is select an image you want to use for style and anther one for the content. Here are some creations of the latest creations the model has generated.

 

All it took for researchers was a mask to bypass iPhone X Face ID

All it took for researchers was a mask to bypass #iPhoneX' #FaceID |  #Security #Privacy #AI

  • Apple Inc. introduced Face ID facial recognition system on September 12, 2017, with the launch of iPhone X and claimed that even high-quality masks such as those used in Hollywood movies couldn’t trick its security system.
  • It turned out Apple was right because all it took for security researchers to fool Face ID was a specially crafted mask, not a Hollywood one.
  • Before going any further, this is what senior vice president of worldwide marketing at Apple Mr. Phil Schiller said about Face ID system during the iPhone X event: – – “Apple engineering teams have even gone and worked with professional mask makers and makeup artists in Hollywood to protect against…
  • More: Anti-surveillance mask enables you to pass as someone else – – However, now, the IT security researchers at Bkav claimed to have bypassed Face ID with a 3D-printed frame, makeup, silicone nose and 2D images along with special processing on the cheeks and around the face, where there are large skin…
  • “So, after nearly 10 years of development, face recognition is not mature enough to guarantee security for computers and smartphones,” said Bkav.

Apple Inc. introduced Face ID facial recognition system on September 12, 2017, with the launch of iPhone X and claimed that even high-quality masks such as
Continue reading “All it took for researchers was a mask to bypass iPhone X Face ID”

Tensorflow Tutorial : Part 2 – Getting Started

Tensorflow Tutorial : Part 2 – Getting Started

  • This post is the second part of the multi-part series on a complete tensorflow tutorial – – – If you have tensorflow already installed, you can just skip to the next section.
  • Below we have the different data types in supported by Tensorflow.
  • Note: Quantitized values [qint8, qint16 and quint8] are special values for tensorflow that help reduce the size of the data.
  • In fact, Google has gone to the extent of introducing Tensorflow Processing Units (TPUs) to speed up computation by leveraging quantitized values – – We will quickly generate some data to get started.
  • In the next part, we will finally be ready to train our first tensorflow model on house prices.

In this multi-part series, we will explore how to get started with tensorflow. This tensorflow tutorial will lay a solid foundation to this popular tool that…
Continue reading “Tensorflow Tutorial : Part 2 – Getting Started”

RFC: Synapse Yellow Paper – Synapseai

  • RFC: Synapse Yellow PaperTL;DR: Yellow paper here.At Synapse we’re building a global decentralized brain that anyone can help build and tap into.Decentralized topologies look alike, whether they’re the internet or the pathways in our brain.When the team originally launched the very first version of our white paper we included everything we were…
  • We ended up with a bunch of feedback that resulted with us creating more of a marketing brochure rather than any technical exposition.We’ve set-out to create that experience inside our yellow paper.
  • The balance between verbosity, succinctness, and technical detail, without being pseudo-technical is what we’re striving for.
  • This is an open, living, and versioned document so expect revisions and upgrades.This is also a Request For Comment (RFC) so we’re looking for feedback from the community on what more we can add or explain, and any contributions anyone would like to add.Best,The Synapse TeamCome visit with us on…

When the team originally launched the very first version of our white paper we included everything we were thinking about, all the questions, and answers that had come up while architecting the…
Continue reading “RFC: Synapse Yellow Paper – Synapseai”

5 ways technology is changing consumer behaviors

5 ways #technology is changing consumer behaviors  #VR #AR #AI

  • In the meantime, some of the major opportunities for brands to engage with consumers is through the creation of high-quality VR and AR content, such as Nike’s VR experience where viewers can feel like they are Neymar juggling a soccer ball with Nike cleats.
  • Simply put, people are now addicted to consuming social media content and brands need to understand how to fit within this new social world.
  • Even if you choose not to use so>cial media content for advertising, take note how it can quickly evolve consumer tastes and how products now fit into social media.
  • This means in the next few years consumers will have a drastic increase in expected quality of customer service from companies and problems will need to be solved effectively.
  • Companies ought to take this time to prepare their consumer relations departments to ensure they have chatbots established on social media, customer helplines if needed, and overall orient customer service as a priority.

While it’s easy to sing the praises of technology, it is important to understand what innovation is significant and what will largely be uninfluential.
Continue reading “5 ways technology is changing consumer behaviors”

The Practical Importance of Feature Selection

The Practical Importance of Feature Selection  #MachineLearning

  • Feature selection is useful on a variety of fronts: it is the best weapon against the Curse of Dimensionality; it can reduce overall training times; and it is a powerful defense against overfitting, increasing generalizability.
  • Feature selection is useful on a variety of fronts: it is the best weapon against the Curse of Dimensionality; it can reduce overall training times; and it is a powerful defense against overfitting, increasing model generalizability.
  • Many times a correct feature selection allows you to develop simpler and faster Machine Learning models.
  • In a time when ample processing power can tempt us to think that feature selection may not be as relevant as it once was, it’s important to remember that this only accounts for one of the numerous benefits of informed feature selection — decreased training times.
  • As Zimbres notes above, with a simple concrete example, feature selection can quite literally mean the difference between valid, generalizable models and a big waste of time.


Feature selection is useful on a variety of fronts: it is the best weapon against the Curse of Dimensionality; it can reduce overall training times; and it is a powerful defense against overfitting, increasing generalizability.

Continue reading “The Practical Importance of Feature Selection”

MarrNet: 3D Shape Reconstruction via 2.5D Sketches

MarrNet: 3D Shape Reconstruction via 2.5D Sketches.  #deeplearning #nips #computervision

  • 3D object reconstruction from a single image is a highly under-determined problem, requiring strong prior knowledge of plausible 3D shapes.
  • In this work, we propose an end-to-end trainable framework, sequentially estimating 2.5D sketches and 3D object shapes.
  • First, compared to full 3D shape, the 2.5D sketch is much easier to be recovered from the 2D image, and to transfer from synthetic to real images.
  • Second, for 3D reconstruction from the 2.5D sketches, we can easily transfer the learned model on synthetic data to real images, as rendered 2.5D sketches are invariant to object appearance variations in real images, including lighting, texture, etc.
  • Third, we derive differentiable projective functions from 3D shapes to 2.5D sketches, making the framework end-to-end trainable on real images, requiring no real-image annotations.

MarrNet
Continue reading “MarrNet: 3D Shape Reconstruction via 2.5D Sketches”