Smart algorithms have taken Google a long way. They helped the company dominate search and create the first software to conquer the complex board game Go. Now the company is betting that algorithms that understand images and text will draw business to its cloud services, make augmented reality popular, and prompt us to search using our smartphone cameras. But some of the algorithms Google is staking its future on aren't equally smart everywhere. The search company's machine learning systems work best on material from a few rich parts of the world, like the US.

We conduct a large-scale study of language models for chord prediction. Specifically, we compare N-gram models to various flavours of recurrent neural networks on a comprehensive dataset comprising all publicly available datasets of annotated chords known to us. This large amount of data allows us to systematically explore hyper-parameter settings for the recurrent neural networks---a crucial step in achieving good results with this model class. Our results show not only a quantitative difference between the models, but also a qualitative one: in contrast to static N-gram models, certain RNN configurations adapt to the songs at test time. This finding constitutes a further step towards the development of chord recognition systems that are more aware of local musical context than what was previously possible.

China has retaliated quickly against proposed United States penalties on Chinese goods and announced 25 percent tariffs on critical American exports, including soya beans, aeroplanes and cars. On Tuesday, the administration of President Donald Trump threatened to slap tariffs on $50bn in Chinese imports across 1,300 categories of products, ranging from industrial robots to locomotives. Beijing's response came hours after the US revealed its plans, with China's foreign ministry saying in a statement that "America's measures to impose tariffs have violated the rules of the World Trade Organisation, and have seriously violated China's legal rights". Soya beans are the top US agricultural export to China and were among the 106 products on which China intends to impose the additional tariffs. The US is the second-biggest soya bean supplier to China, after Brazil.

Are you an IoT / M2M professional? Find out more about Global IoT Market Trends and IoT Companies & IoT Startups. At the Telecom Infra Project (TIP) Summit, Telefónica S.A.and Facebook announced a joint collaboration to reduce the digital divide in Latin America (LatAm) by targeting challenges to rural […]

The "spaghetti code" is a pejorative term to refer to computer programs that have a complex and incomprehensible flow control structure. Its name derives from the fact that this type of code seems to resemble a plate of spaghetti, that is, a pile of intricate and knotted threads. Traditionally this style of programming is usually associated with basic and ancient languages, where the flow was controlled by very primitive control statements such as GO TO and using line numbers. An example of language that invited the use of spaghetti code is Microsoft's QBasic in its first versions. Throughout these decades programming has been evolving, from spaghetti code to functional programming and from functional programming to object-oriented programming with modularity, abstraction, encapsulation, decoupling capacity.

The field of multi-agent system (MAS) is an active area of research within artificial intelligence, with an increasingly important impact in industrial and other real-world applications. In a MAS, autonomous agents interact to pursue personal interests and/or to achieve common objectives. Distributed Constraint Optimization Problems (DCOPs) have emerged as a prominent agent model to govern the agents' autonomous behavior, where both algorithms and communication models are driven by the structure of the specific problem. During the last decade, several extensions to the DCOP model have been proposed to enable support of MAS in complex, real-time, and uncertain environments. This survey provides an overview of the DCOP model, offering a classification of its multiple extensions and addressing both resolution methods and applications that find a natural mapping within each class of DCOPs. The proposed classification suggests several future perspectives for DCOP extensions and identifies challenges in the design of efficient resolution algorithms, possibly through the adaptation of strategies from different areas.

Machine-learning algorithms have gained popularity in recent years in the field of ecological modeling due to their promising results in predictive performance of classification problems. While the application of such algorithms has been highly simplified in the last years due to their well-documented integration in commonly used statistical programming languages such as R, there are several practical challenges in the field of ecological modeling related to unbiased performance estimation, optimization of algorithms using hyperparameter tuning and spatial autocorrelation. We address these issues in the comparison of several widely used machine-learning algorithms such as Boosted Regression Trees (BRT), k-Nearest Neighbor (WKNN), Random Forest (RF) and Support Vector Machine (SVM) to traditional parametric algorithms such as logistic regression (GLM) and semi-parametric ones like generalized additive models (GAM). Different nested cross-validation methods including hyperparameter tuning methods are used to evaluate model performances with the aim to receive bias-reduced performance estimates. As a case study the spatial distribution of forest disease Diplodia sapinea in the Basque Country in Spain is investigated using common environmental variables such as temperature, precipitation, soil or lithology as predictors. Results show that GAM and RF (mean AUROC estimates 0.708 and 0.699) outperform all other methods in predictive accuracy. The effect of hyperparameter tuning saturates at around 50 iterations for this data set. The AUROC differences between the bias-reduced (spatial cross-validation) and overoptimistic (non-spatial cross-validation) performance estimates of the GAM and RF are 0.167 (24%) and 0.213 (30%), respectively. It is recommended to also use spatial partitioning for cross-validation hyperparameter tuning of spatial data.

It is midnight on January 18, 2017, and the Outbrain Click Prediction machine learning competition has just finished. It has been three and a half months of working late. As I scroll through the leaderboard page, I found my name in the 19th position, which was the top 2% from nearly 1,000 competitors. Not bad for the first Kaggle competition I had decided to put a real effort in! One of the reasons why I managed to score well was the fact that Google Cloud Platform (GCP) made my life easier and I could focus on the data.

Geoglyphs in the southern Amazon are evidence of a once-thriving population. Before Spanish invaders conquered South America, sparse groups of nomadic people clustered around the Amazon River, leaving the surrounding rain forest pristine and untouched. New research suggests a very different story--an Amazonian region peppered with rain forest villages, ceremonial earthworks, and a much larger population than previously thought. The research, funded in part by the National Geographic Society and published today in the journal Nature Communications, challenges a common perception of the pre-Columbian Amazon rain forest as sparsely populated. That perception has endured despite 16th-century accounts of large, interconnected villages that go against modern assumptions.

The recent successes of AI have captured the wildest imagination of both the scientific communities and the general public. Robotics and AI amplify human potentials, increase productivity and are moving from simple reasoning towards human-like cognitive abilities. Current AI technologies are used in a set area of applications, ranging from healthcare, manufacturing, transport, energy, to financial services, banking, advertising, management consulting and government agencies. The global AI market is around 260 billion USD in 2016 and it is estimated to exceed 3 trillion by 2024. To understand the impact of AI, it is important to draw lessons from it's past successes and failures and this white paper provides a comprehensive explanation of the evolution of AI, its current status and future directions.