When customer experience or operations teams within communications service providers (CSPs) want to adopt artificial intelligence (AI), they first must find a way to'sell' it to finance executives, and perhaps more importantly to the employees who could be displaced by the technology. Neither is an easy job, according to a panel of experts gathered here at Action Week in Dallas. Jerrid Hamann, Digital Customer Experience Strategist, Verizon, who has worked for the company for about a year, spoke about his experience at another telco where he was trying to implement AI for customer service. "I was trying to convince finance that we needed the new tools to improve customer experience," he said. "Either they were very skeptical and suspicious saying it sounds like science fiction and is not something we want to invest our money in…or at the other end of the spectrum they say, 'Oh wow, we can save that much money? Let's lay off the entire contact center'. When you get that kind of reaction you have to dial it back and explain that it's something that has to be phased in."

So, you've hunkered down and finally completed that online course on machine learning. Now, you have all sorts of ideas running through your mind on developing your own intelligent code and neural networks. You assume you'll have to fork out a considerable wedge for a decent GPU-powered number-crunching rig, because your handy lightweight laptop is not going to cut it during the intensive network training process. So, seeing as you'll dabble with this on and off initially, you're looking at renting out GPUs on cloud. Your heart drops a little when you total up the cloud instance costs.

Paul Allen has made a name for himself as a co-founder of Microsoft, a supporter of artificial intelligence research and a contributor to causes such as wildlife conservation -- so it only makes sense that the Seattle-area billionaire wants to use machine learning to further his philanthropic goals. His latest contribution comes through the Seattle-based Vulcan Machine Learning Center for Impact, or VMLCI. "Its mission will be to apply the tools of machine learning and AI for good," Bill Hilf, CEO of Paul Allen's Vulcan Inc., said today in a tweet. VMLCI's strategy meshes with the mission of the Allen Institute for Artificial Intelligence, whose motto is "AI for the Common Good." The center aims to forge collaborative partnerships with corporations, academic institutions and other organizations to help connect folks working on social and environmental causes with the machine-learning resources they need.

When discussing ways of ensuring European competitiveness in the age of artificial intelligence, we often talk about encouraging and incentivizing existing European companies to start utilizing artificial intelligence. This is very important challenge to tackle for Europe to remain competitive and an issue that has spurred a variety of activities in Finland as well. AI can be a significant competitive advantage for companies that adopt it early, take AI to the core of their business and commit to it. While AI can deliver great results in terms of e.g. In many sectors, small businesses can challenge large traditional companies using new types of artificial intelligence solutions. These solutions not only improve the quality of services and reduce costs but also create completely new industries and services.

Machine learning has undoubtedly been hugely successful in driving progress in AI, but it implicitly brings with it the train-test evaluation paradigm. This standard evaluation only encourages behavior that is good on average; it does not ensure robustness as demonstrated by adversarial examples, and it breaks down for tasks such as dialogue that are interactive or do not have a correct answer. In this talk, I will describe alternative evaluation paradigms with a focus on natural language understanding tasks, and discuss ramifications for guiding progress in AI in meaningful directions. Percy Liang is an Assistant Professor of Computer Science at Stanford University (B.S. from MIT, 2004; Ph.D. from UC Berkeley, 2011). His research spans machine learning and natural language processing, with the goal of developing trustworthy agents that can communicate effectively with people and improve over time through interaction.

The machine I built costs $3k and has the parts shown below. There's one 1080 Ti GPU to start (you can just as easily use the new 2080 Ti for Machine Learning at $500 more -- just be careful to get one with a blower fan design), a 12 Core CPU, 64GB RAM, and 1TB M.2 SSD. You can add three more GPUs easily for a total of four. Assuming your 1 GPU machine depreciates to $0 in 3 years (very conservative), the chart below shows that if you use it for up to 1 year, it'll be 10x cheaper, including costs for electricity. Amazon discounts pricing if you have a multi-year contract, so the advantage is 4–6x for multi-year contracts.

Online class imbalance learning constitutes a new problem and an emerging research topic that focusses on the challenges of online learning under class imbalance and concept drift. Class imbalance deals with data streams that have very skewed distributions while concept drift deals with changes in the class imbalance status. Little work exists that addresses these challenges and in this paper we introduce queue-based resampling, a novel algorithm that successfully addresses the co-existence of class imbalance and concept drift. The central idea of the proposed resampling algorithm is to selectively include in the training set a subset of the examples that appeared in the past. Results on two popular benchmark datasets demonstrate the effectiveness of queue-based resampling over state-of-the-art methods in terms of learning speed and quality.

Non-convex optimization with global convergence guarantees is gaining significant interest in machine learning research in recent years. However, while most works consider either offline settings in which all data is given beforehand, or simple online stochastic i.i.d. settings, very little is known about non-convex optimization for adversarial online learning settings. In this paper we focus on the problem of Online Principal Component Analysis in the regret minimization framework. For this problem, all existing regret minimization algorithms are based on a positive semidefinite convex relaxation, and hence require quadratic memory and SVD computation (either thin of full) on each iteration, which amounts to at least quadratic runtime per iteration. This is in stark contrast to a corresponding stochastic i.i.d. variant of the problem which admits very efficient gradient ascent algorithms that work directly on the natural non-convex formulation of the problem, and hence require only linear memory and linear runtime per iteration. This raises the question: \textit{can non-convex online gradient ascent algorithms be shown to minimize regret in online adversarial settings?} In this paper we take a step forward towards answering this question. We introduce an \textit{adversarially-perturbed spiked-covariance model} in which, each data point is assumed to follow a fixed stochastic distribution, but is then perturbed by adversarial noise. We show that in a certain regime of parameters, when the non-convex online gradient ascent algorithm is initialized with a "warm-start" vector, it provably minimizes the regret with high probability. We further discuss the possibility of computing such a "warm-start" vector. Our theoretical findings are supported by empirical experiments on both synthetic and real-world data.

Much attention has been devoted recently to the generalization puzzle in deep learning: large, deep networks can generalize well, but existing theories bounding generalization error are exceedingly loose, and thus cannot explain this striking performance. Furthermore, a major hope is that knowledge may transfer across tasks, so that multi-task learning can improve generalization on individual tasks. However we lack analytic theories that can quantitatively predict how the degree of knowledge transfer depends on the relationship between the tasks. We develop an analytic theory of the nonlinear dynamics of generalization in deep linear networks, both within and across tasks. In particular, our theory provides analytic solutions to the training and testing error of deep networks as a function of training time, number of examples, network size and initialization, and the task structure and SNR. Our theory reveals that deep networks progressively learn the most important task structure first, so that generalization error at the early stopping time primarily depends on task structure and is independent of network size. This suggests any tight bound on generalization error must take into account task structure, and explains observations about real data being learned faster than random data. Intriguingly our theory also reveals the existence of a learning algorithm that proveably out-performs neural network training through gradient descent. Finally, for transfer learning, our theory reveals that knowledge transfer depends sensitively, but computably, on the SNRs and input feature alignments of pairs of tasks.

Standard artificial neural networks suffer from the well-known issue of catastrophic forgetting, making continual or lifelong learning problematic. Recently, numerous methods have been proposed for continual learning, but due to differences in evaluation protocols it is difficult to directly compare their performance. To enable more meaningful comparisons, we identified three distinct continual learning scenarios based on whether task identity is known and, if it is not, whether it needs to be inferred. Performing the split and permuted MNIST task protocols according to each of these scenarios, we found that regularization-based approaches (e.g., elastic weight consolidation) failed when task identity needed to be inferred. In contrast, generative replay combined with distillation (i.e., using class probabilities as "soft targets") achieved superior performance in all three scenarios. In addition, we reduced the computational cost of generative replay by integrating the generative model into the main model by equipping it with generative feedback connections. This Replay-through-Feedback approach substantially shortened training time with no or negligible loss in performance. We believe this to be an important first step towards making the powerful technique of generative replay scalable to real-world continual learning applications.