The Reverend Kate Bottley, possibly the most famous vicar in Britain thanks to her regular appearances on Gogglebox, seems to have a god-given gift for the 1983 arcade game Track & Field. She has just thrashed the Guardian's games editor Keith Stuart at the classic button-bashing sports sim. She takes the 100m sprint with ease, before breaking a record in the javelin. "That's it, I'm retiring undefeated," she declares to the small crowd that's gathered around her at the GameCity festival taking place at Nottingham's National Video Arcade (NVA). She walks away from the machine.

In this talk we will present some techniques that we use on a day to day basis in our research, where we combine our internet-wide data scanning and acquisition platform with ML/Data science techniques which allows us to find things faster or extract results in a more automated way. We will focus on practical cases and examples that even our audience at home will be able to use if they want. A couple of examples we will look at is how to classify images such as VNC screenshots, we will look at network scans and using machine learning to classify them and also the use of natural language processing to analyze CVEs. We will also talk a bit about a data analysis and classification pipeline architecture, we will look at the different technologies and what they do and how they can be used. We will start by giving a very brief entry to the data science world and talk about: Technologies Techniques How these relate to infosec Algorithms and how they can be used How people can come into the world of data and machine learning Data visualization techniques and what are the best choices for different types of data A couple of examples we will look at is how to classify images such as VNC or x11 screenshots, OCR, we will look at network scans and using machine learning to classify them and also the use of natural language processing to analyze CVEs.

It is widely believed that the prediction accuracy of decision tree models is invariant under any strictly monotone transformation of the individual predictor variables. However, this statement may be false when predicting new observations with values that were not seen in the training-set and are close to the location of the split point of a tree rule. The sensitivity of the prediction error to the split point interpolation is high when the split point of the tree is estimated based on very few observations, reaching 9% misclassification error when only 10 observations are used for constructing a split, and shrinking to 1% when relying on 100 observations. This study compares the performance of alternative methods for split point interpolation and concludes that the best choice is taking the mid-point between the two closest points to the split point of the tree. Furthermore, if the (continuous) distribution of the predictor variable is known, then using its probability integral for transforming the variable ("quantile transformation") will reduce the model's interpolation error by up to about a half on average. Accordingly, this study provides guidelines for both developers and users of decision tree models (including bagging and random forest).

We investigate if kernel regularization methods can achieve minimax convergence rates over a source condition regularity assumption for the target function. These questions have been considered in past literature, but only under specific assumptions about the decay, typically polynomial, of the spectrum of the the kernel mapping covariance operator. In the perspective of distribution-free results, we investigate this issue under much weaker assumption on the eigenvalue decay, allowing for more complex behavior that can reflect different structure of the data at different scales.

Hedge funds have been in the doldrums and face mounting pressure to justify their fees. Will artificial intelligence come to the rescue? A growing number of hedge funds are putting money behind the idea that a branch of AI called machine learning could provide a way to get back on top. A software program that searches for patterns in more data than even the most sleep-deprived junior analyst could examine, and then tests its hypotheses against even more data. What can satellite shots of mall parking lots tell you when combined with in-store sales data?

Lip reading is a way of understanding speech by interpreting a person's lip movement. However, human speech is highly complex and nuanced, where one lip movement could correspond to different phonemes, or basic units of sound. Therefore, the practice is prone to errors, which can sometimes lead to humorous results. Scientists from Oxford University have described an artificial intelligence system, called LipNet, which can accurately read lips. The system employs deep learning to train itself using 29,000 three-second-long videos labeled with captions.

Artificial intelligence (AI), or the process by which computers or robots perform tasks that need human intelligence, was one of the key themes of this week's Web Summit in the Portuguese capital, Lisbon. The poll, conducted among 224 venture capitalists attending the conference showed 53 percent believed AI would destroy millions of jobs and 93 percent saw governments as unprepared for this. The survey also found that 83 percent of the investors canvassed expected Britain's exit from the European Union to damage Europe's economy and 77 percent believed it would damage British startups. London is widely seen as the main tech startup hub in Europe, thanks to its large pool of talent and a much bigger pool of funding than in rival centers. Cities like Berlin, Amsterdam and Lisbon are eager to attract more tech startups.

SAP uncovered its machine learning arsenal at this week's TechEd event in Barcelona. In an approach that is similar to Microsoft's, it is infusing machine learning across its business applications and making services available to encourage partners to build the techniques into their HANA built applications. Wherever you look there appears to be a machine learning element. There is a newly developed SAP Machine Learning Platform which will be available to partners and developers next year. The new HANA 2 features analytics improvements with new processing engines for text, spatial, graph and streaming data, driven by newly added classification, association, time series and regression machine learning algorithms.

Back in the 90s, I used to design automated systems to help pilots fly civil aircraft. Today, this would probably be called designing artificial intelligence (AI) or smart systems or something similar. One of the big questions for us back then was deciding how much responsibility to give to the aircraft to fly, manage failures and generally make and execute decisions, as well as how much to leave to the pilots. One of the hot issues was whether the pilot or smart systems should have the final call when things got tough. Controversially, in several cases, authority was given to the aircraft because the aircraft could manage specific situations better.

We've improved OpenSfM--the technology we use to create 3D reconstructions from images. By adding post-processing, we get denser 3D point clouds resulting in better visualization, positioning, and much more. At Mapillary we build and use OpenSfM to find the relative positions of images and create smooth transitions between them. That process is called Structure from Motion. It works by matching a few thousand points between images, and then figuring out the 3D positions of those points as well as the positions of the cameras simultaneously.