There is no doubt that Big Data has been one of the most popular topics among marketers and tech enthusiasts for several years. Within the big data domain one of the most promising fields is deep learning which has evolved into one of tech's most exciting and promising disciplines in the field of AI (Artificial Intelligence). The popularity of deep learning peaked in March 2016, when Google's DeepMind AI program called AlphaGo bested Lee Sedol, the celebrated player of the board game "Go", by winning four out of five games. After the match it was revealed that a relatively new AI technique called "deep learning" was responsible for the victory. According to scientists, deep learning technology has the potential to transform the entire AI area.

The top machine learning videos on YouTube include lecture series from Stanford and Caltech, Google Tech Talks on deep learning, using machine learning to play Mario and Hearthstone, and detecting NHL goals from live streams.

IBM (NYSE: IBM) and NVIDIA announced that they had produced the first server optimized for NVIDIA's newest GPUs. This partnership would accelerate the training of deep learning neural networks faster than the traditional design. This was achieved by connecting the GPUs directly to the servers' processors using a special high-speed link that reduces bottlenecks and increases processing speed. Additionally, they announced a deep learning software toolkit, called IBM PowerAI, which would run on the optimized server. This combination provides an easy-to-deploy system for enterprise data and research scientists who are involved in AI research.

The project combines the power of both memorization and generalization, to better reflect the properties that make the human brain such an effective learning machine. They provide an in-depth example that illustrates the project's purpose and potential via a fictional food delivery app. The Harvard NLP and Visual Computing groups announce LSTMVis, a visual analysis tool for recurrent neural networks (RNNs). RNNs learn a black-box hidden state representation, and changes in these representations are challenging to study. The tool makes it easier to visually observe and isolate patterns in state changes.

We have been seeing a sudden rise in the deployment of Artificial Intelligence (AI), Machine Learning (ML), and Deep Learning (DL). It looks like the long "AI winter" is finally over. It is interesting to note that AI was mentioned by Alan Turing in a paper he wrote back in 1950 to suggest that there is possibility to build machines with true intelligence. Then in 1956, John McCarthy organized a conference at Dartmounth and coined the phrase Artificial Intelligence. Much of the next three decades did not see much activity and hence the phrase "AI Winter" was coined.

With artificial intelligence all the rage these days, market trackers are attempting to gauge just where the technology is headed and which industry sectors will lead development for specific big data and other enterprise use cases. The latest attempt comes from Evans Data Corp. in the form of an AI and big data survey released on Wednesday (Jan. The survey of 440 AI developers found that more than one-third of respondents are focusing on deep learning techniques, with most targeting the financial and insurance sectors. Other sectors where deep learning implementations are expected to have an impact include the Internet of Things (14.9 percent) and "non-computer" manufacturing (12.5 percent), reported the market researcher based in Santa Cruz, Calif. Nearly one-third of AI developers focused on deep learning implementations are relying on numerical inputs as the most common data type, Evans Data added.

In a landmark achievement for artificial intelligence, a poker bot developed by researchers in Canada and the Czech Republic has defeated several professional players in one-on-one games of no-limit Texas hold'em poker. Perhaps most interestingly, the academics behind the work say their program overcame its human opponents by using an approximation approach that they compare to "gut feeling." "If correct, this is indeed a significant advance in game-playing AI," says Michael Wellman, a professor at the University of Michigan who specializes in game theory and AI. "First, it achieves a major milestone (beating poker professionals) in a game of prominent interest. Second, it brings together several novel ideas, which together support an exciting approach for imperfect-information games."

Recurrent neural networks with various types of hidden units have been used to solve a diverse range of problems involving sequence data. Two of the most recent proposals, gated recurrent units (GRU) and minimal gated units (MGU), have shown comparable promising results on example public datasets. In this paper, we introduce three model variants of the minimal gated unit (MGU) which further simplify that design by reducing the number of parameters in the forget-gate dynamic equation. These three model variants, referred to simply as MGU1, MGU2, and MGU3, were tested on sequences generated from the MNIST dataset and from the Reuters Newswire Topics (RNT) dataset. The new models have shown similar accuracy to the MGU model while using fewer parameters and thus lowering training expense. One model variant, namely MGU2, performed better than MGU on the datasets considered, and thus may be used as an alternate to MGU or GRU in recurrent neural networks.

The standard LSTM recurrent neural networks while very powerful in long-range dependency sequence applications have highly complex structure and relatively large (adaptive) parameters. In this work, we present empirical comparison between the standard LSTM recurrent neural network architecture and three new parameter-reduced variants obtained by eliminating combinations of the input signal, bias, and hidden unit signals from individual gating signals. The experiments on two sequence datasets show that the three new variants, called simply as LSTM1, LSTM2, and LSTM3, can achieve comparable performance to the standard LSTM model with less (adaptive) parameters.

We present QuickNet, a fast and accurate network architecture that is both faster and significantly more accurate than other fast deep architectures like SqueezeNet. Furthermore, it uses less parameters than previous networks, making it more memory efficient. We do this by making two major modifications to the reference Darknet model (Redmon et al, 2015): 1) The use of depthwise separable convolutions and 2) The use of parametric rectified linear units. We make the observation that parametric rectified linear units are computationally equivalent to leaky rectified linear units at test time and the observation that separable convolutions can be interpreted as a compressed Inception network (Chollet, 2016). Using these observations, we derive a network architecture, which we call QuickNet, that is both faster and more accurate than previous models. Our architecture provides at least four major advantages: (1) A smaller model size, which is more tenable on memory constrained systems; (2) A significantly faster network which is more tenable on computationally constrained systems; (3) A high accuracy of 95.7 percent on the CIFAR-10 Dataset which outperforms all but one result published so far, although we note that our works are orthogonal approaches and can be combined (4) Orthogonality to previous model compression approaches allowing for further speed gains to be realized.