If you are a newcomer to the Deep Learning area, the first question you may have is "Which paper should I start reading from?" After reading above papers, you will have a basic understanding of the Deep Learning history, the basic architectures of Deep Learning model(including CNN, RNN, LSTM) and how deep learning can be applied to image and speech recognition issues. The following papers will take you in-depth understanding of the Deep Learning method, Deep Learning in different areas of application and the frontiers. "Deep compression: Compressing deep neural network with pruning, trained quantization and huffman coding."

IBM's virtual health care assistant powered by Watson, which has memorized most of the world's medical journals and texts, is an example of this. Enterprise is already feeling the impact of AI assistants like IBM's Watson health care tool, but Gartner says these systems will become more "conversational." Powered by machine learning and AI, such systems will learn how to adapt to humans and vice versa. Gartner believes five major focal points will enable the new capabilities and business models of digital business, including information systems, customer experience, analytics and intelligence, the IoT, and business ecosystems.

Traffic congestion costs the U.S. economy 121 billion a year, mostly due to lost productivity, and produces about 25 billion kilograms of carbon dioxide emissions, Carnegie Mellon University professor of robotics Stephen Smith told the audience at a White House Frontiers Conference last week. In urban areas, drivers spend 40 percent of their time idling in traffic, he added. The next step is to have traffic signals talk to cars. Pittsburgh is the test bed for Uber's self-driving cars, and Smith's work on AI-enhanced traffic signals that talk with self-driving cars is paving the way for the ultimately fluid and efficient autonomous intersections.

It's home to Google Assistant, the new conversational assistant that Google plans to bring to many more of its consumer products. The first sign that Allo is a bit smarter than the standard messaging app is through a feature called Smart Replies. Even more impressive, Google has combined this feature with its photo recognition abilities, so the app is able to suggest responses to photos that are shared within your conversation. Things that Google can do "natively" like conversions, simple Google searches and translations will appear directly in your chat while other queries will surface links to Google results.

The Associated Press and Graphiq, which specializes in using artificial intelligence to rapidly create interactive data-driven infographics, announced a new partnership Tuesday. The partnership, whose financials were not disclosed, will make it easier for the AP to provide data infographics for its stories while exposing Graphiq to a worldwide user audience. "Today we're reaching hundreds of millions of readers a month, but the AP reaches half the world's population every day," Alex Rosenberg, Graphiq vice president, told Inc. Rosenberg noted that Graphiq will embed some of its staffers in AP newsrooms to be readily available throughout the news outlet's reporting process. The AP will also make Graphiq's entire visualization catalog--the company creates thousands of new visualizations each week--readily available to news wires' countless clients.

In a significant step forward for artificial intelligence, Alphabet's hybrid system -- called a Differential Neural Computer (DNC) -- uses the existing data storage capacity of conventional computers while pairing it with smart AI and a neural net capable of quickly parsing it. "These models can learn from examples like neural networks, but they can also store complex data like computers," wrote DeepMind researchers Alexander Graves and Greg Wayne. Much like the brain, the neural network uses an interconnected series of nodes to stimulate specific centers needed to complete a task. Instead of having to learn every possible outcome to find a solution, DeepMind can derive an answer from prior experience, unearthing the answer from its internal memory rather than from outside conditioning and programming.

Blighty's Science and Technology Committee released Robotics and Artificial Intelligence, while the White House delivered Preparing for the Future of Artificial Intelligence and National Artificial Intelligence Research and Development Strategic Plan. Limitations to AI research were also recognised in the separate report, The National Artificial Intelligence Research and Development Strategic Plan. The US plans to develop shared public data sets for AI training and testing. Bryson makes a good point as the White House report was timed just before Obama was due to host the White House's Frontiers Conference.

Professor Stephen Hawking on Wednesday opened a new artificial intelligence research centre at Britain's Cambridge University. "Alongside the benefits, AI will also bring dangers, like powerful autonomous weapons, or new ways for the few to oppress the many. But catastrophic scenarios aside, the development of AI, which allows robots to execute almost all human tasks, directly threatens millions of jobs. "It's about how to ensure intelligent artificial systems have goals aligned with human values" and ensure computers don't evolve spontaneously in "new, unwelcome directions", Cave said.

Then we use another map transformation, which will apply the ParseObs function to transform each Array of Double in the RDD into an Array of Cancer Observation objects. The toDF() method transforms the RDD of Array[[Cancer Observation]] into a Dataframe with the Cancer Observation class schema. Below the data is split into a training data set and a test data set, 70% of the data is used to train the model, and 30% will be used for testing. In this blog post, we showed you how to get started using Apache Spark's machine learning Logistic Regression for classification.

With traditional software engineering techniques we always define functions to solve problems. The idea is that the neural network finds the right weights for each neuron to eventually learn how to calculate the expected output. But even better, there's a library called Keras, also written in Python, which creates a higher level API and uses TensorFlow as it's backend. Let's teach a neural network to understand the XOR gate.