Artificial Intelligence (AI), long the subject of science fiction, is now becoming more and more widespread and is seen as an increasingly important computer science across multiple industries. In Financial Services in particular, Machine Learning and Natural Language Processing is increasingly used today to make sense of big, complex data in a wide range of areas. One such area is regulatory compliance. The use of AI – particularly Natural Language Understanding (NLU), a subset of Natural Language Processing – can help firms to realise a number of benefits, including improving the speed and efficiency with which they achieve compliance, and making that compliance much more robust. As we've seen over just the last couple of years with the introduction of MiFID I & II, UCITS, AIFMD and the like, there is a constant stream of documents being issued by regulators, which can each run to hundreds, or even thousands, of pages.

Disruption in the workforce is hardly a new phenomenon. Mechanisation of manufacturing, mass production and the advent of the internet and computers have all changed the way that work is done. Earlier waves of industrialisation have primarily affected low-skilled manual labour and past improvements in technology have typically made jobs at the lower end of the skills spectrum obsolete – for example, flight navigators but not pilots; typists but not data analysts. There is wide acceptance that this has led to productivity improvements and higher economic growth – new jobs were generated that led to improvements in standards of living. The benefits have overwhelmingly outweighed the costs and there has never been a better time to be a human being. The current wave, characterised by automation becoming smarter, machine-to-machine communication, artificial intelligence and continued technological improvements – and otherwise described at the fourth industrial revolution – still brings uncertainty and threatens a broader range of occupations and skill levels.

You may have seen our previous posts on machine learning -- specifically, how to let your code learn from text and working with stop words, stemming, and spam. So today, we're going to build our machine learning-based spam filter, using the tools we walked through in those posts: tokenizer, stemmer, and naive bayes classifier. We are going to work with bluebird promise library here, so if you are not used to promises, please take a look at the bluebird API reference. Before we begin, it's important to have good training data. You can download some here -- we are interested in two.

Over the past year or two, I've heard these buzz words being tossed around a lot, and it's something that has definitely seized my curiosity recently. Deep learning is an area of active research these days, and if you've kept up with the field of computer science, I'm sure you've come across at least some of these terms at least once. Deep learning can be an indimidating concept, but it's becoming increasingly important these days. Moreover, deep learning methods are beating out traditional machine learning approaches on virtually every single metric. So what exactly is deep learning? And most importantly, why should you even care? If you're new to computer science, and you've followed me up till this point, please stick with me. Certain optional sections of this article may get a little math heavy (marked with a *), but I want to make this subject accessible to everyone, computer science major or not.

Baidu, the world's No. 3 force in online advertising after Google and Facebook, sees artificial intelligence as the next big thing. And it's using AI to power advertising, voice recognition technology and self-driving cars. During its annual conference Thursday in Beijing, the search giant announced news on the autonomous vehicle front, saying it had won clearance from the California Department of Motor Vehicles to test its cars there. It also signed a partnership with U.S. chip-maker Nvidia to work on self-driving cars together. Despite all the investment on futuristic projects, Baidu got nearly 93% of its revenue in the last quarter from online marketing.

Most deep learning techniques are extensions or adaptions of ANN's, called deep nets. Different configurations of deep nets are suitable for different machine learning tasks: Restricted Boltzman Machines (RBM's) (Smolensky 1986; Hinton & Salakhutdinov, 2006) and Autoencoders (Vincent, Larochelle, Bengio, & Manzagol,2008) are the main deep learning techniques for finding patterns in unlabeled data. This includes tasks such as feature extraction, pattern recognition and other unsupervised learning settings.

Forget about Minority Report and its sexy gesture interface – predicting the future is very different from what you see in the movies. Applying the right statistical models allows you to gain insights from the information at your disposal. The hidden patterns unveiled by the process makes it possible to make predictions. This is what we call predictive analytics. This is how the retail industry is able to predict what customers buy according to the time of the month or other items they have just purchased.

There are lots of writings about how data analytics changes business and everything, but deep down it only matters how you put these analytics in action. This is a software engineering perspective. Following the success of AlphaGo, the machine learning software that defeated the Go master Lee Sedol, the Wired magazine released an article titled "The End of Code" with a provoking highlight This is a very interesting transition. In reality though, we are not going to train whole large systems like this, but more atomic functions. You see, training an ERP system like a dog would be one heck of an agility track.

What differentiates us from machines is not our ability to think, but our ability to reason and to make cross connections between two unrelated phenomenons. There is no doubt that machines have been getting smarter and now, there are concerns that they may take over work from humans. So, while Percy Spencer, inventor of the microwave, could figure out that micro waves from an active radar could be used for heating food, a machine would have just dismissed the accident as a black swan event. Although, there is increasing automation--especially for jobs that entail repetitive tasks, for instance, self-driving cars--there is still time before artificial evolution allows machines to turn our lives into a sci-fi movie gone horribly wrong. in that context, we need to address the question whether these are really as intelligent as their human masters. In order to delve into that question, one needs to understand artificial intelligence and its evolution.

We demonstrate a hybrid machine learning method to classify schizophrenia patients and healthy controls, using functional magnetic resonance imaging (fMRI) and single nucleotide polymorphism (SNP) data. The method consists of four stages: (1) SNPs with the most discriminating information between the healthy controls and schizophrenia patients are selected to construct a support vector machine ensemble (SNP-SVME). The method was evaluated by a fully validated leave-one-out method using 40 subjects (20 patients and 20 controls). The classification accuracy was: 0.74 for SNP-SVME, 0.82 for Voxel-SVME, 0.83 for ICA-SVMC, and 0.87 for Combined SNP-fMRI. Experimental results show that better classification accuracy was achieved by combining genetic and fMRI data than using either alone, indicating that genetic and brain function representing different, but partially complementary aspects, of schizophrenia etiopathology.