Machine learning employs dynamical algorithms that mimic the human capacity to learn, where the reinforcement learning ones are among the most similar to humans in this respect. On the other hand, adaptability is an essential aspect to perform any task efficiently in a changing environment, and it is fundamental for many purposes, such as natural selection. Here, we propose an algorithm based on successive measurements to adapt one quantum state to a reference unknown state, in the sense of achieving maximum overlap. The protocol naturally provides many identical copies of the reference state, such that in each measurement iteration more information about it is obtained. In our protocol, we consider a system composed of three parts, the "environment" system, which provides the reference state copies; the register, which is an auxiliary subsystem that interacts with the environment to acquire information from it; and the agent, which corresponds to the quantum state that is adapted by digital feedback with input corresponding to the outcome of the measurements on the register. With this proposal we can achieve an average fidelity between the environment and the agent of more than $90\% $ with less than $30$ iterations of the protocol. In addition, we extend the formalism to $ d $-dimensional states, reaching an average fidelity of around $80\% $ in less than $400$ iterations for $d=$ 11, for a variety of genuinely quantum as well as semiclassical states. This work paves the way for the development of quantum reinforcement learning protocols using quantum data, and the future deployment of semi-autonomous quantum systems.

ONNEGATIVE matrix factorization finds its application in the fields of machine learning and in connection with inverse problems, mostly. It became immensely popular after Lee and Seung derived multiplicative update rules that made the up until then additive steps in the direction of the negative gradient obsolete [1]. In [2], they gave empirical evidence of their convergence to a stationary point, using (a) the squared Euclidean distance, and, (b) the generalized Kullback-Leibler divergence as the contrast function. The factorization's origins can be traced back to [3], [4]. To better deal with noisy data, the notion of a basis is (most commonly) abandoned in favor of an overcomplete frame or dictionary, and sparsity becomes a desired property of the coefficient matrix.

We study the problem of maximizing a monotone set function subject to a cardinality constraint $k$ in the setting where some number of elements $\tau$ is deleted from the returned set. The focus of this work is on the worst-case adversarial setting. While there exist constant-factor guarantees when the function is submodular, there are no guarantees for non-submodular objectives. In this work, we present a new algorithm Oblivious-Greedy and prove the first constant-factor approximation guarantees for a wider class of non-submodular objectives. The obtained theoretical bounds are the first constant-factor bounds that also hold in the linear regime, i.e. when the number of deletions $\tau$ is linear in $k$. Our bounds depend on established parameters such as the submodularity ratio and some novel ones such as the inverse curvature. We bound these parameters for two important objectives including support selection and variance reduction. Finally, we numerically demonstrate the robust performance of Oblivious-Greedy for these two objectives on various datasets.

Binarization of digital documents is the task of classifying each pixel in an image of the document as belonging to the background (parchment/paper) or foreground (text/ink). Historical documents are often subject to degradations, that make the task challenging. In the current work a deep neural network architecture is proposed that combines a fully convolutional network with an unrolled primal-dual network that can be trained end-to-end in order to achieve state of the art binarization on four out of seven datasets. Document binarization is formulated as a energy minimization problem. A fully convolutional neural network is trained for semantic labeling of pixels to provide class labeling cost associated with each pixel. This cost estimate is refined along the edges to compensate for any over or under estimation of the under represented fore-ground class using a primal-dual approach. We provide necessary overview on proximal operator that facilitates theoretical underpinning in order to train a primal-dual network using a gradient descent algorithm. Numerical instabilities encountered due to the recurrent nature of primal-dual approach are handled. We provide experimental results on document binarization competition dataset along with network changes and hyperparameter tuning required for stability and performance of the network. The network when pre-trained on synthetic dataset performs better as per the competition metrics.

We propose a simple extension to the ReLU-family of activation functions that allows them to shift the mean activation across a layer towards zero. Combined with proper weight initialization, this alleviates the need for normalization layers. We explore the training of deep vanilla recurrent neural networks (RNNs) with up to 144 layers, and show that bipolar activation functions help learning in this setting. On the Penn Treebank and Text8 language modeling tasks we obtain competitive results, improving on the best reported results for non-gated networks. In experiments with convolutional neural networks without batch normalization, we find that bipolar activations produce a faster drop in training error, and results in a lower test error on the CIFAR-10 classification task.

Deep convolutional neural networks (CNNs) used in practice employ potentially hundreds of layers and $10$,$000$s of nodes. Such network sizes entail significant computational complexity due to the large number of convolutions that need to be carried out; in addition, a large number of parameters needs to be learned and stored. Very deep and wide CNNs may therefore not be well suited to applications operating under severe resource constraints as is the case, e.g., in low-power embedded and mobile platforms. This paper aims at understanding the impact of CNN topology, specifically depth and width, on the network's feature extraction capabilities. We address this question for the class of scattering networks that employ either Weyl-Heisenberg filters or wavelets, the modulus non-linearity, and no pooling. The exponential feature map energy decay results in Wiatowski et al., 2017, are generalized to $\mathcal{O}(a^{-N})$, where an arbitrary decay factor $a>1$ can be realized through suitable choice of the Weyl-Heisenberg prototype function or the mother wavelet. We then show how networks of fixed (possibly small) depth $N$ can be designed to guarantee that $((1-\varepsilon)\cdot 100)\%$ of the input signal's energy are contained in the feature vector. Based on the notion of operationally significant nodes, we characterize, partly rigorously and partly heuristically, the topology-reducing effects of (effectively) band-limited input signals, band-limited filters, and feature map symmetries. Finally, for networks based on Weyl-Heisenberg filters, we determine the prototype function bandwidth that minimizes---for fixed network depth $N$---the average number of operationally significant nodes per layer.

When human settlers one day reach distant worlds, they might take refuge inside extensive underground networks known as planetary caves. Scientists in recent years have proposed volcanic caves as a potential place for safe habitats on the moon and Mars. But, to understand how this might work, we must first look at similar systems here at home. A team of ESA-backed researchers recently ventured into the La Cueva de los Verdes lava tube in Spain to map the nearly 5 mile (8 kilometer) 'lava tube' in unprecedented detail, revealing the intricate features of the'volcanic wormhole.' The team used what's known as the point cloud technique to map La Cueva de Los Verdes in unprecedented detail, capturing millions of measurements with lasers and cameras.

Artificial intelligence (AI) in banking is a fast-developing reality as banks around the world are looking to leverage the technology to reduce costs and create better client experiences. "Based on our UBS Evidence Lab survey of 86 banks, an optimal scenario of limited disruption suggests AI technology could potentially lead to a 3.4% revenue uplift and cost savings of 3.9% over the next three years," UBS strategist Philip Finch wrote a recent note titled Is AI the next revolution in retail banking? Goldman Sachs estimates a £26 billion (US$36.2 billion) to £33 billion (US$46 billion) in annual "cost savings and new revenue opportunities" within the financial sector by 2025, enabled by AI and machine learning. One of the most basic objectives for AI at banks is a reduction in time "wasted" on any task that can be automated. These include rules-based tasks, such as entry, validation, and manipulation of data, as well as creation, uploading, and exporting of data files.

Have you ever booked a flight or hotel and then noticed that the prices drop after you've shelled out? A computerised legal tool has expanded its offerings in a bid to help travellers recuperate losses like this on holiday bookings. DoNotPay - masterminded by Stanford University tech whizz Joshua Browder - uses a'robot lawyer' to find legal loopholes and negotiate cheaper prices or re-book reservations for customers if prices lower after they've paid. DoNotPay - masterminded by Stanford University tech whizz Joshua Browder - uses a'robot lawyer' to find legal loopholes and negotiate cheaper prices or re-book travel reservations for customers if prices lower after they've paid Once a customer signs up, the sophisticated bot automatically looks for all the travel confirmations in the user's email inbox and checks for fluctuations in price about 17,000 times a day until the departure date. If the flight or hotel price drops, DoNotPay claims it would bag the better deal and make the vendor refund the difference.

POLITICO Establishes Global AI Forum for Business Leaders and Policymakers with Accenture as Founding Partner First in series of AI summits will take place on March 19-20, 2018 in Brussels, Belgium BRUSSELS; March 8, 2018 – POLITICO, with Accenture Applied Intelligence as the founding partner, is launching a global artificial intelligence (AI) forum to help business leaders and government policymakers understand the impact of AI innovation and to inform responsible use of AI. As AI-based decisions have an increasing impact on human lives, the initiative aims to empower decision-makers to build a framework for governance in pivotal and unchartered territory. The initiative will hold a series of AI summits and roundtables in multiple cities in Europe and the US. The first AI Summit will take place on March 19-20, 2018 in Brussels, Belgium. Carlos Moedas, European Union Commissioner for Research Science and Innovation, and John Delaney, US Congressman and founder of the bipartisan AI Caucus, are two keynote participants of the event.