Manifold learning and dimensionality reduction techniques are ubiquitous in science and engineering, but can be computationally expensive procedures when applied to large data sets or when similarities are expensive to compute. To date, little work has been done to investigate the tradeoff between computational resources and the quality of learned representations. We present both theoretical and experimental explorations of this question. In particular, we consider Laplacian eigenmaps embeddings based on a kernel matrix, and explore how the embeddings behave when this kernel matrix is corrupted by occlusion and noise. Our main theoretical result shows that under modest noise and occlusion assumptions, we can (with high probability) recover a good approximation to the Laplacian eigenmaps embedding based on the uncorrupted kernel matrix. Our results also show how regularization can aid this approximation. Experimentally, we explore the effects of noise and occlusion on Laplacian eigenmaps embeddings of two real-world data sets, one from speech processing and one from neuroscience, as well as a synthetic data set.

Most learning algorithms are not invariant to the scale of the function that is being approximated. We propose to adaptively normalize the targets used in learning. This is useful in value-based reinforcement learning, where the magnitude of appropriate value approximations can change over time when we update the policy of behavior. Our main motivation is prior work on learning to play Atari games, where the rewards were all clipped to a predetermined range. This clipping facilitates learning across many different games with a single learning algorithm, but a clipped reward function can result in qualitatively different behavior. Using the adaptive normalization we can remove this domain-specific heuristic without diminishing overall performance.

This thesis uses deep learning algorithms to forecast financial data. The deep learning framework is used to train a neural network. The deep neural network is a Deep Belief Network (DBN) coupled to a Multilayer Perceptron (MLP). It is used to choose stocks to form portfolios. The portfolios have better returns than the median of the stocks forming the list. The stocks forming the S&P 500 are included in the study. The results obtained from the deep neural network are compared to benchmarks from a logistic regression network, a multilayer perceptron and a naive benchmark. The results obtained from the deep neural network are better and more stable than the benchmarks. The findings support that deep learning methods will find their way in finance due to their reliability and good performance.

Lonnie Johnson was brought up in Mobile Alabama in the 1960s, when black children were not expected to go far, but such was his talent for engineering that he worked for Nasa, and helped test the first stealth bomber. But as he explains here, the invention that made his fortune was a water pistol - the extremely powerful Super Soaker. It started with my dad. He gave me my first lesson in electricity, explaining that it takes two wires for electric current to flow - one for the electrons to go in, the other for them to come out. And he showed me how to repair irons and lamps and things like that. The kids in the neighbourhood took to calling me "the Professor".

Robert Gordon has written a magnificent book on the economic history of the United States over the last one and a half centuries. His study focuses on what he calls the "special century" from 1870 to 1970--in which living standards increased more rapidly than at any time before or after. The book is without peer in providing a statistical analysis of the uneven pace of growth and technological change, in describing the technologies that led to the remarkable progress during the special century, and in concluding with a provocative hypothesis that the future is unlikely to bring anything approaching the economic gains of the earlier period. The message of Rise and Fall is this. For most of human history, economic progress moved at a crawl. According to the economic historian Bradford DeLong, from the first rock tools used by humanoids three million years ago, to the earliest cities ten thousand years ago, through the Middle Ages, to the beginning of the Industrial Revolution around 1800, living standards doubled (with a growth of 0.00002 percent per year). Another doubling took place over the subsequent period to 1870. Then, according to standard calculations, the world economy took off. Gordon focuses on growth in the United States.

These claims are not ludicrous on their own. We've seen rapid advancements in technology over the past decades; we know computers are growing more powerful and more accessible by the month. Already in 2011, a supercomputer named Watson won a game of Jeopardy against two former champions, using a mixture of AI and all-important natural-language processing. The future is here and it may soon outstrip us. Kurzweil's timeline of the technological singularity is based on the law of accelerating returns, wherein the more powerful computers become, the faster they advance.

We hear from Apple CEO Tim Cook during the company's quarterly earnings calls, but lately the leader of Apple has been opening up in ways that non-analysts can understand. First he sat down with Fast Company, and on Monday an in-depth Washington Post interview on Cook's 5-year anniversary as CEO offered more insight into the company's past failures and future surprises. A lot has changed in the last five years, Cook told the Post. Steve Jobs passed away just six weeks after Cook took the reins, which came as a shock, he said. Apple expanded its product lineup after Jobs's death, and some of those devices were conceived of and developed entirely without Jobs's input.

Estimates suggest that by the year 2020 we will have 20 billion devices connected to the Internet. Industries are shifting, lines are blurring, and markets are changing all as a result of a technology and digital-centric approach to commerce and business. In this podcast, Cliff Justice, Partner in KPMG's Innovation and Enterprise Solutions group, sits down with Stan Lepeak to discuss: The convergence of robotic process automation (RPA), machine learning, cognitive computing, artificial intelligence, and advanced analytics are driving unparalleled business model transformation. To learn more visit KPMG's Digital Labor website: www.kpmg.com/us/digitallabor. To discover more, read Embracing the Cognitive Era.

From a furtive Chinese character inked on a shoulder to a full arm sleeve, tattoos have enjoyed a revival in popularity in recent years. But soon they could do more than simply decorate the skin. Researchers have created temporary tattoos that can are sensitive to touch and can be used to control electronic devices like smartphones. Working in collaboration with Microsoft Research, researchers at the Massachusetts Institute of Technology's (MIT) Media Lab produced the temporary tattoos, whichare called DuoSkin. Using gold leaf, the temporary transfers can be used as touch sensitive trackpads or to transmit information to a smartphone.

Nasa has announced that it has found evidence of flowing water on Mars. Scientists have long speculated that Recurring Slope Lineae -- or dark patches -- on Mars were made up of briny water but the new findings prove that those patches are caused by liquid water, which it has established by finding hydrated salts. Several hundred camped outside the London store in Covent Garden. The 6s will have new features like a vastly improved camera and a pressure-sensitive "3D Touch" display