Machine Learning’s Mediocre Gains

Hedge funds using AI suffer the same shortcomings as their more traditional peers

  • Hedge funds using vast amounts of data, computing power, and machine-learning techniques to make money are drawing investors’ attention.
  • But their brief track records show they suffer the same shortcomings as their more traditional peers.
  • The Eurekahedge AI Hedge Fund Index, which tracks 12 of these money pools, has outperformed hedge fund peers since 2013 but failed to beat the SP 500 Index.

Like hedge funds, AI strategies have struggled to beat the stock market

Continue reading “Machine Learning’s Mediocre Gains”

Google creates a neural network that can carry out EIGHT different tasks at once in a step towards making AI behave more like a human

Google creates a #neuralnetwork that can carry out EIGHT tasks at once  #AI

  • But now, for the first time, Google has created a system that can do eight tasks at once, including image and speech recognition and language translation.
  • The inspiration for the MultiModel comes from how the brain transforms sensory input from modalities such as sound, vision and taste, and transforms them into a single shared representation and back out as language or actions.
  • ‘It can detect objects in images, provide captions, recognise speech, translate between four pairs of languages, and do grammatical constituency parsing at the same time’, the researchers, led by Lukasz Kaiser wrote in their blog.
  • ‘When designing MultiModel it became clear that certain elements from each domain of research (vision, language and audio) were integral to the model’s success in related tasks’, the researchers wrote.
  • ‘To our surprise, this happens even if the tasks come from different domains that would appear to have little in common, e.g., an image recognition task can improve performance on a language task’, the researchers said.

The MultiModel system has been created by researchers from Mountain View, California-based Google Brain. The system  can be taught to do eight tasks at once.
Continue reading “Google creates a neural network that can carry out EIGHT different tasks at once in a step towards making AI behave more like a human”

GitHub

Our implementation of graph auto-encoders (in TensorFlow) is now available on GitHub:

  • This is a TensorFlow implementation of the (Variational) Graph Auto-Encoder model as described in our paper:

    Graph Auto-Encoders (GAEs) are end-to-end trainable neural network models for unsupervised learning, clustering and link prediction on graphs.

  • GAEs have successfully been used for:

    GAEs are based on Graph Convolutional Networks (GCNs), a recent class of models for end-to-end (semi-)supervised learning on graphs:

    A high-level introduction is given in our blog post:

    In order to use your own data, you have to provide

    Have a look at the function in for an example.

  • In this example, we load citation network data (Cora, Citeseer or Pubmed).
  • The original datasets can be found here: and here (in a different format): can specify a dataset as follows:

    You can choose between the following models:

    Please cite our paper if you use this code in your own work:

gae – Implementation of Graph Auto-Encoders in TensorFlow
Continue reading “GitHub”

10 Super Scary Developments In Artificial Intelligence – Viral FreQ

  • That is, unfortunately, a very real fear these days, as mankind is making more and more advances in the science and technology behind AI components and robotics.
  • All it takes is for one robot to grow p*ssed off before they decide to form an uprising against humanity.
  • Lying and cheating is actually a normal human behavior.
  • But robots are only just learning to do this all on their own.
  • Researchers from the Georgia Institute of Technology developed a series of robots capable of cheating and deception towards others.

Stephen Hawking, Bill Gates, and even Elon Musk are considered some of the most brilliant men on the planet, but they are all absolutely terrified of artificial intelligence taking over the world one day. That is, unfortunately, a very real fear these days, as mankind is making more and more advances in the science and technology behind AI components and robotics. All it takes is for one robot to grow p*ssed off before they decide to form an uprising against humanity. We should all be terrified of that happening. Here are ten truly scary developments in AI.
Continue reading “10 Super Scary Developments In Artificial Intelligence – Viral FreQ”

The future of flying cars? Scientists create drones that can fly, drive and communicate with each other to avoid a collision

Scientists have created drones that can fly and drive  #drone #AI #uav

  • To make them capable of driving, the team put two small motors with wheels on the bottom of each drone.
  • Adding the driving component to the drone slightly reduced its battery life, meaning that the maximum distance it could fly decreased 14 per cent to about 91 metres.
  • To make them capable of driving, the team put two small motors with wheels on the bottom of each drone.
  • Adding the driving component to the drone slightly reduced its battery life, meaning that the maximum distance it could fly decreased by 14 per cent to about 91 metres (300ft).
  • ‘As we begin to develop planning and control algorithms for flying cars, we are encouraged by the possibility of creating robots with these capabilities at small scale,’ CSAIL Director Daniela Rus said.

Researchers from MIT’s Computer Science and Artificial Intelligence Laboratory in Massachusetts say the drone could lead to machines that fly into disaster zones.
Continue reading “The future of flying cars? Scientists create drones that can fly, drive and communicate with each other to avoid a collision”

Start watching Alex Castrounis’s class on Skillshare.

Join - Skillshare

  • : Skillshare uses Javascript for some of its core functionality.
  • It is highly recommended that you turn on Javascript in your preferences and reload the page.

Join a global learning community to create, connect, and collaborate with students around the globe. Skillshare offers online classes to fit your schedule with bite-sized lessons from industry leaders that will help you turn ideas into action.
Continue reading “Start watching Alex Castrounis’s class on Skillshare.”

Microsoft is using 400 million PCs to build antivirus protection

Microsoft is using 400 million PCs to build antivirus protection

  • To prevent the next global malware crisis, an upcoming update will rely on machine learning from more than 400 million computers running Windows 10, Microsoft said Tuesday.
  • In its Fall Creators Update, Microsoft will use a wide range of data coming from its cloud programs such as Azure, Endpoint and Office to create an artificial intelligence antivirus that can pick up on malware behavior, said Rob Lefferts, director of program management for Windows Enterprise and Security.
  • If new malware is detected on any computer running Windows 10 in the world, Microsoft said it will be able to develop a signature for it and protect all the other users worldwide.
  • Microsoft sees artificial intelligence as the next solution for security as attacks get more sophisticated.
  • Using cloud data from Microsoft Office to develop malware signatures is crucial, for example, because recent attacks relied on Word vulnerabilities.

An upcoming security update will incorporate machine learning from millions of computers fending off malware, the company says.
Continue reading “Microsoft is using 400 million PCs to build antivirus protection”

Wimbledon serves up a new take on journalism in the age of AI

#ai Wimbledon serves up a new take on journalism in the age of AI  via @thenextweb

  • But that’s not to say that it’s stuck in the past — thanks to a partnership with IBM, Wimbledon is using Watson to automate key parts of the iconic tennis event.
  • But this year round, the job is placed in the hands of the Watson AI.
  • Watson can generate highlight packages without any human input.
  • One IBM representative said that eventually, the company will be able to reduce this time to just 30 minutes

    Watson is also tasked with tagging photos.

  • But surely the argument could be made that Watson is essentially automating entry-level journalism jobs?

Wimbledon is 150 years old. That’s not to say it’s stuck in the past.
Continue reading “Wimbledon serves up a new take on journalism in the age of AI”

Fake news is your fault, this startup is trying to help

Fake news is your fault, this startup is trying to help

  • In Ghulati’s opinion, every part of the chain — from journalists to politicians, platforms to media organizations — needs to improve to combat fake news.
  • That’s why Factmata, which is backed by the Google Digital News Initiative, is trying create tools that help people to become their own fact-checkers — making it easy to attain factual data about news stories you come across.
  • Meaning that when you open a news article, Factmata’s main tool would provide factual context in real time.
  • That means that Factmata’s tool won’t actively weed out fake news.
  • However, Ghulati states that fake news can never be solved solely by democratizing fact-checking — it will always need experts to also be a part of the system.

Readers are the key group when it comes to fighting fake news. That’s why a UK startup is creating a system that helps people to fact-check news they read.
Continue reading “Fake news is your fault, this startup is trying to help”

A step-by-step guide to building a simple chess AI – freeCodeCamp

A step-by-step guide to building a simple #chess #AI using #Javascript

  • The simplest way to achieve this is to count the relative strength of the pieces on the board using the following table:With the evaluation function, we’re able to create an algorithm that chooses the move that gives the highest evaluation:The only tangible improvement is that our algorithm will now capture a piece if it can.Black plays with the aid of the simple evaluation function.
  • This is done by using the Minimax algorithm.In this algorithm, the recursive tree of all possible moves is explored to a given depth, and the position is evaluated at the ending “leaves” of the tree.After that, we return either the smallest or the largest value of the child to the parent node, depending on whether it’s a white or black to move.
  • The best move for white is b2-c3, because we can guarantee that we can get to a position where the evaluation is -50With minimax in place, our algorithm is starting to understand some basic tactics of chess:Minimax with depth level 2.
  • This helps us evaluate the minimax search tree much deeper, while using the same resources.The alpha-beta pruning is based on the situation where we can stop evaluating a part of the search tree if we find a move that leads to a worse situation than a previously discovered move.The alpha-beta pruning does not influence the outcome of the minimax algorithm — it only makes it faster.The alpha-beta algorithm also is more efficient if we happen to visit first those paths that lead to good moves.The positions we do not need to explore if alpha-beta pruning isused and the tree is visited in the described order.With alpha-beta, we get a significant boost to the minimax algorithm, as is shown in the following example:The number of positions that are required to evaluate if we want to perform a search with depth of 4 and the “root” position is the one that is shown.Follow this link to try the alpha-beta improved version of the chess AI.Step 5: Improved evaluation functionThe initial evaluation function is quite naive as we only count the material that is found on the board.
  • We can decrease or increase the evaluation, depending on the location of the piece.With the following improvement, we start to get an algorithm that plays some “decent” chess, at least from the viewpoint of a casual player:Improved evaluation and alpha-beta pruning with search depth of 3.

At each step, we’ll improve our algorithm with one of these time-tested chess-programming techniques. I’ll demonstrate how each affects the algorithm’s playing style. We’ll use the chess.js library…
Continue reading “A step-by-step guide to building a simple chess AI – freeCodeCamp”