As many as 30pc of existing roles in the UK could be automated by 2030 with the most at risk industries being waste management, transportation and manufacturing, according to an analysis by PwC. However, the report stressed that automation won't result in rocketing unemployment. "The UK employment rate is at its highest level now since comparable records began in 1971, despite advances in digital and other labour-saving technologies," said John Hawksworth, chief economist at PwC. Mr Hawksworth anticipates that manual and routine tasks will be susceptible to automation, with social skills and creative roles being more protected. "That said, no industry is entirely immune from future advances in robotics and AI," he said.

Bayesian inference for complex models is challenging due to the need to explore high-dimensional spaces and multimodality and standard Monte Carlo samplers can have difficulties effectively exploring the posterior. We introduce a general purpose rejection-free ensemble Markov Chain Monte Carlo (MCMC) technique to improve on existing poorly mixing samplers. This is achieved by combining parallel tempering and an auxiliary variable move to exchange information between the chains. We demonstrate this ensemble MCMC scheme on Bayesian inference in Factorial Hidden Markov Models. This high-dimensional inference problem is difficult due to the exponentially sized latent variable space. Existing sampling approaches mix slowly and can get trapped in local modes. We show that the performance of these samplers is improved by our rejection-free ensemble technique and that the method is attractive and "easy-to-use" since no parameter tuning is required.

The nearest neighbor (NN) classifier endowed with the dynamic time warping (DTW) distance is one of the most popular methods in time series classification [9, 44]. Application examples include electrocardiogram frame classification [16], gesture recognition [2, 32], speech recognition [24], and voice recognition [23]. Two disadvantages of the naive NN method are high storage and computation requirements. Storage requirements are high, because the entire training set needs to be retained for being able to execute its classification rule. Computation requirements are high, because classifying a test example demands calculation of DTW distances between the test and all training examples.

We consider additive models built with trend filtering, i.e., additive models whose components are each regularized by the (discrete) total variation of their $(k+1)$st (discrete) derivative, for a chosen integer $k \geq 0$. This results in $k$th degree piecewise polynomial components, (e.g., $k=0$ gives piecewise constant components, $k=1$ gives piecewise linear, $k=2$ gives piecewise quadratic, etc.). In univariate nonparametric regression, the localized nature of the total variation regularizer used by trend filtering has been shown to produce estimates with superior local adaptivity to those from smoothing splines (and linear smoothers, more generally) (Tibshirani [2014]). Further, the structured nature of this regularizer has been shown to lead to highly efficient computational routines for trend filtering (Kim et al. [2009], Ramdas and Tibshirani [2016]). In this paper, we argue that both of these properties carry over to the additive models setting. We derive fast error rates for additive trend filtering estimates, and prove that these rates are minimax optimal when the underlying function is itself additive and has component functions whose derivatives are of bounded variation. We show that such rates are unattainable by additive smoothing splines (and by additive models built from linear smoothers, in general). We argue that backfitting provides an efficient algorithm for additive trend filtering, as it is built around the fast univariate trend filtering solvers; moreover, we describe a modified backfitting procedure whose iterations can be run in parallel. Finally, we conduct experiments to examine the empirical properties of additive trend filtering, and outline some possible extensions.

A recurring problem faced when training neural networks is that there is typically not enough data to maximize the generalization capability of deep neural networks(DNN). There are many techniques to address this, including data augmentation, dropout, and transfer learning. In this paper, we introduce an additional method which we call Smart Augmentation and we show how to use it to increase the accuracy and reduce overfitting on a target network. Smart Augmentation works by creating a network that learns how to generate augmented data during the training process of a target network in a way that reduces that networks loss. This allows us to learn augmentations that minimize the error of that network. Smart Augmentation has shown the potential to increase accuracy by demonstrably significant measures on all datasets tested. In addition, it has shown potential to achieve similar or improved performance levels with significantly smaller network sizes in a number of tested cases.

Inverse reinforcement learning (IRL) has become a useful tool for learning behavioral models from demonstration data. However, IRL remains mostly unexplored for multi-agent systems. In this paper, we show how the principle of IRL can be extended to homogeneous large-scale problems, inspired by the collective swarming behavior of natural systems. In particular, we make the following contributions to the field: 1) We introduce the swarMDP framework, a sub-class of decentralized partially observable Markov decision processes endowed with a swarm characterization. 2) Exploiting the inherent homogeneity of this framework, we reduce the resulting multi-agent IRL problem to a single-agent one by proving that the agent-specific value functions in this model coincide. 3) To solve the corresponding control problem, we propose a novel heterogeneous learning scheme that is particularly tailored to the swarm setting. Results on two example systems demonstrate that our framework is able to produce meaningful local reward models from which we can replicate the observed global system dynamics.

Someday soon, you'll be able to tell Siri to book a flight to Chicago, get you a hotel on the Miracle Mile, reserve a table at Morton's, and get you tickets for the Cubbies. Today, to accomplish the same task, you need to dive into five or six different apps for hotels, flights, dinner reservations, sports tickets, and transfers between all the locations, and spend maybe 30 to 60 minutes sweating the details. But the day might be coming sooner rather than later that Siri can do it all for you. Today, Apple announced that it has acquired Workflow, an innovative iOS app that glues together functionality from multiple apps into a single, simple ... flow. Workflow currently allows you to take pictures and automate editing of them, or enter data and flow it to multiple places, or make PDFs out of web pages, or just about anything else you can imagine.

Artificial Intelligence/Machine Learning field is getting a lot of attention right now, and knowing where to start can be a little difficult. I've been dabbling in this field, so I thought of curating the best resources in one place. All of these are curated based on if it's an inspiring read or a valuable resource. I hope this curated list help you get started on what you need to know about AI/Machine Learning on a technical level. Design intelligent agents to solve real-world problems including, search, games, machine learning, logic, and constraint satisfaction problems.

Disrupting the streaming music market is extremely difficult to do, given the entrenched positions of Pandora and Spotify in the world market. But Brazilian company SuperPlayer is doing just that, with its streaming music service that is simple, curated, and effectively halves the price of Spotify's Premium service in Brazil. With the ability to listen to music offline without consuming data, Spotify should be worried. But the most impressive disruption that SuperPlayer has pulled off is using chatbots in their influencer marketing schemes. It's no secret that chatbots are essential to customer service, but SuperPlayer is the first to integrate music into chatbots, and effectively.

YouTube has long had an automatic captioning system that, thanks to Google's machine learning advances in recent years, has gotten pretty good at automatically transcribing spoken words in a video. As the company announced today, its technology is now able to take this a step further by also captioning some of the ambient sounds like [LAUGHTER], [APPLAUSE] and [MUSIC]. For now, the automatic effects captioning is actually restricted to those exactly these three sounds. The reason for this, Google says, is due to the fact that these are also exactly the sounds that most video producers manually caption right now. "While the sound space is obviously far richer and provides even more contextually relevant information than these three classes, the semantic information conveyed by these sound effects in the caption track is relatively unambiguous, as opposed to sounds like [RING] which raises the question of "what was it that rang – a bell, an alarm, a phone?,"