Cristina is a malware researcher at Bitdefender's Antimalware Labs. For the past 6 years, she has demonstrated strong expertise in reverse engineering, exploit analysis, threat analysis and automated systems. She has a graduate degree in Computer Science from the "Gheorghe Asachi" Technical University in Iasi and is now pursuing a PhD in Machine Learning theory in malware detection systems.

The firm queried 138 U.S. tech CEOs, and 95 percent expect to increase the size of their workforces over the next three years. Some 55 percent expect to grow at least 6 percent. This growth in tech jobs for humans might have been even larger, however, but some positions will be filled by automation and machine learning systems: about three quarters of the tech CEOs expect automation and machine learning tools to replace at least 5 percent of their sales, marketing, technology, and manufacturing workforces. "The majority of technology companies plan to increase their human workforce at least 6 percent over the next three years while adding cognitive systems to create a new class of digital labor that can enhance human skills and expertise." Does this mean tech companies are going to start issuing reports on their digital labor force?

Elon Musk spoke at MIT for an event marking the 100th anniversary of its Department of Aeronautics and Astronautics. During the Q&A session, Elon warned of the civilisation threatening dangers of Artificial Intelligence if not contained.

The next BriefingsDirect business innovation thought leadership discussion focuses on how new modes of buying and evaluating goods and services are disrupting business procurement. We'll hear now from a leading industry analyst on how machine learning, cloud services, and artificial intelligence-enabled human agents are all combining to change the way that companies can order services, buy goods, and even hire employees and contractors. This business process innovation exchange comes to you in conjunction with the Tradeshift Innovation Day held in New York on June 22, 2016. To learn more about how new trends are driving innovation into invoicing and spend management, please join me in welcoming Pierre Mitchell, Chief Research Officer and Managing Director at Azul Partners, where he leads the Spend Matters Procurement research activities. The discussion is moderated by me, Dana Gardner, Principal Analyst at Interarbor Solutions.

Recently, we have seen artificial intelligence triumph over humans in Jeopardy and chess. And there is a growing presence of virtual assistants like Alexa, Cortana, and Siri that populate our computers, phones, and homes. It's only a matter of time before A.I.-powered assistants play a significant role in the workplace, experts say. In fact, the global intelligent virtual assistant market is forecast to be worth 5.1 billion by 2022, up from – 600 million in 2014, according to Transparency Market Research. What are the potential benefits and challenges of giving smart virtual assistants a home in the enterprise?

One of the biggest misconceptions about Google's self-driving car right now is that you can't pull up Google Maps, pick a destination and tell the car to go there. That's because to learn new routes, the car has to be "trained" by a human driver at least once or twice first. But now a number of organizations, including Ford Motor Company, Stanford University and an investment firm run by Yahoo co-founder Jerry Yang, have invested 6.6 million into a company that promises to leapfrog that navigation issue by creating cheap, detailed maps that driverless cars will be able to read on the fly. And these maps will be created by regular drivers such as yourself, according to Civil Maps, the company behind the idea. In that respect, the concept is a bit like another Google-owned product, Waze.

Video Friday is your weekly selection of awesome robotics videos, collected by your melodramatic Automaton bloggers. We'll also be posting a weekly calendar of upcoming robotics events for the next two months; here's what we have so far (send us your events!): Let us know if you have suggestions for next week, and enjoy today's videos. I'm not entirely sure why being "targeted" by a bird robot like this would appeal to kids, but here you go: Fun fact: this works with real seagulls too. "This is a demonstration of Engineered Art's telepresence app on its Robothespian Humanoid robot. Although recorded multiple times to capture different camera angles, what you are seeing is a real conversation with Robothespian. It is being controlled and voiced remotely over the internet. There is no trickery here. Although we scripted some of the conversation to ensure it was repeatable, none of the content was pre-programmed. It was all created in real time using a headset and the telepresence app. It is an excellent way to ensure genuine human interaction, without being tripped up with it's complexities."

Which authors of this paper are endorsers? Disable MathJax (What is MathJax?)

Machine learning becomes more and more popular, and there are now many demonstrations available over the internet which help to demonstrate some ideas about algorithms in a more vivid way. There are certain arguments against having courses overloaded with interactive things since those frequently prevent students from diving into mathematics beyond machine learning, but generally I consider demonstrations helpful. ROC curve is fairly simple subject, but having a demo is nice way to demonstrate some important limit cases. I have prepared simple html demo for this. There are also other interesting demonstrations like t-SNE and RNNs, you're welcome to check Andrej's github page.

A grand challenge in machine learning is the development of computational algorithms that match or outperform humans in perceptual inference tasks such as visual object and speech recognition. The key factor complicating such tasks is the presence of numerous nuisance variables, for instance, the unknown object position, orientation, and scale in object recognition or the unknown voice pronunciation, pitch, and speed in speech recognition. Recently, a new breed of deep learning algorithms have emerged for high-nuisance inference tasks; they are constructed from many layers of alternating linear and nonlinear processing units and are trained using large-scale algorithms and massive amounts of training data. The recent success of deep learning systems is impressive – they now routinely yield pattern recognition systems with nearor super-human capabilities – but a fundamental question remains: Why do they work? Intuitions abound, but a coherent framework for understanding, analyzing, and synthesizing deep learning architectures has remained elusive.