Bayesian learning of model parameters in neural networks is important in scenarios where estimates with well-calibrated uncertainty are important. In this paper, we propose Bayesian quantized networks (BQNs), quantized neural networks (QNNs) for which we learn a posterior distribution over their discrete parameters. We provide a set of efficient algorithms for learning and prediction in BQNs without the need to sample from their parameters or activations, which not only allows for differentiable learning in QNNs, but also reduces the variance in gradients. We demonstrate BQNs achieve both lower predictive errors and better-calibrated uncertainties than E-QNN (with less than 20% of the negative log-likelihood). A Bayesian approach to deep learning considers the network's parameters to be random variables and seeks to infer their posterior distribution given the training data. Models trained this way, called Bayesian neural networks (BNNs) (Wang & Y eung, 2016), in principle have well-calibrated uncertainties when they make predictions, which is important in scenarios such as active learning and reinforcement learning (Gal, 2016). Furthermore, the posterior distribution over the model parameters provides valuable information for evaluation and compression of neural networks. There are three main challenges in using BNNs: (1) Intractable posterior: Computing and storing the exact posterior distribution over the network weights is intractable due to the complexity and high-dimensionality of deep networks. These challenges are typically addressed either by making simplifying assumptions about the distributions of the parameters and activations, or by using sampling-based approaches, which are expensive and unreliable (likely to overestimate the uncertainties in predictions). Our goal is to propose a sampling-free method which uses probabilistic propagation to deterministically learn BNNs. A seemingly unrelated area of deep learning research is that of quantized neural networks (QNNs), which offer advantages of computational and memory efficiency compared to continuous-valued models.

In this article, we focus on the analysis of the potential factors driving the spread of influenza, and possible policies to mitigate the adverse effects of the disease. To be precise, we first invoke discrete Fourier transform (DFT) to conclude a yearly periodic regional structure in the influenza activity, thus safely restricting ourselves to the analysis of the yearly influenza behavior. Then we collect a massive number of possible region-wise indicators contributing to the influenza mortality, such as consumption, immunization, sanitation, water quality, and other indicators from external data, with $1170$ dimensions in total. We extract significant features from the high dimensional indicators using a combination of data analysis techniques, including matrix completion, support vector machines (SVM), autoencoders, and principal component analysis (PCA). Furthermore, we model the international flow of migration and trade as a convolution on regional influenza activity, and solve the deconvolution problem as higher-order perturbations to the linear regression, thus separating regional and international factors related to the influenza mortality. Finally, both the original model and the perturbed model are tested on regional examples, as validations of our models. Pertaining to the policy, we make a proposal based on the connectivity data along with the previously extracted significant features to alleviate the impact of influenza, as well as efficiently propagate and carry out the policies. We conclude that environmental features and economic features are of significance to the influenza mortality. The model can be easily adapted to model other types of infectious diseases.

A new study suggest that ability to convey kindness through facial expressions may have been a key factor in human evolution. The study was conducted by Matteo Zanella and a team of researchers at the University of Milan, and published this week in Science Advances. The team compared genetic data from human stem cells with samples from the remains of two Neanderthals and one Denisovan, a sister species to Neanderthals found in central Asia. They specifically focused on the BAZ1B gene, which has been connected to Williams-Beuren syndrome, a condition that causes people to develop wide mouths and small noses that give a generally kind and welcoming impression. The BAZ1B gene has also been associated with the evolution of two extra muscles in dogs that allow them to widen and narrow their eyes in expressive ways, something wolves aren't able to do.

We think of AI as an arbiter of neutrality, but when fed biased data it churns out biased results. At the beginning of 2017, Amazon's machine learning division shuttered an artificial intelligence (AI) project it had been working on for the past three years. A team in its machine learning wing had been building computer programmes designed to review job applicants' resumes, giving them star-ratings from one to five – not unlike the way shoppers can rate products purchased from Amazon online. However, within a year of the project beginning, the company realised its system was biased against female applicants. The software was trained to vet applicants by observing patterns in resumes submitted to the company over a ten-year period, the majority of which – due to the male-dominance of the tech industry – came from men.

In late September, Beijing unveiled to the world Daxing, a glimmering $11 billion airport showcasing technologies such as robots and facial recognition scanners that many other airports worldwide are either adopting or are now considering. Daxing fits the description of what experts hail as a "smart airport." Just as a smart home is where internet-connected devices control functions like security and thermostats, smart airports use cloud-based technologies to simplify and improve services. Of course, many of the nearly 4,000 scheduled service airports across the world are still embarrassingly antiquated. The good news for aviation is that more facilities are investing, finally, to better serve airlines, suppliers, and travelers. This year, airports worldwide will spend $11.8 billion -- 68 percent more than the level three years ago -- on information technology, according to an estimate published this month by SITA (Société Internationale de Telecommunications Aeronautiques, an airline-owned tech provider). A few trends are driving the rise of smart airports. Flight volumes are increasing, so airports need better ways to process flyers. Airports need better ways to make money, too, by encouraging passengers to spend more in their shops and restaurants. Data is growing in importance. Everything happening at an airport, from where passengers are flowing to which items are selling in stores, generates data. Airports can analyze this data to spot opportunities for eking out fatter profits. They can sell the data to third-parties as well.

Take a look around and you will find a lot of self-proclaimed marketing professionals. However, there are only a few names who have actually made it to the top and gotten their art acknowledged. Today on Branex Talks, we are privileged to have such a gentleman with us. To date, he has helped businesses and startups founders scale their business. He has extensive experience working with various professionals in 70 countries, including Taiwan, Spain, Ireland, the US and the UK.

With the Essential2020 plan all but complete, Orange has released the details of the Engage2025 strategy to drive growth over the next five years. The new strategy is going to be focused on four key pillars; reinventing the operator business model, accelerating growth in the developing markets and emerging segments, integrate artificial intelligence at the centre of every aspect of the business, and building sustainability goals through the organization. "If I had to summarise Engage2025, Orange's new strategic plan, I would use two words: growth and sustainability," said CEO Stephane Ricard. "The first one is growth. We are going to grow our core business – connectivity – by adding to our competitive edge and by making the most of our network infrastructure. We are also going to foster growth beyond connectivity in Europe thanks to three elements which set us apart from our competitors, namely Africa & the Middle East, B2B IT services and financial services. At Orange we are convinced that in the years ahead strong economic performance will not be possible without exemplary performance on social and environmental issues."

We consider the example of a deployment of an air pollution monitoring network in Kampala, an East African city. Air pollution contributes to over three million deaths globally each year(Lelieveld and others, 2015). Kampala has one of the highest concentrations of fine particulate matter (PM 2.5) of any African city Mead (2017) Hence we know little about its distribution or extent. Lower cost devices do exist, but these do not, on their own, provide the accuracy required for decision makers. In our case study, the Kampala network of sensors consists largely of low cost optical particle counters (OPCs) that give estimates of the PM2.5 particulate concentration.

Statistical learning theory provides bounds of the generalization gap, using in particular the Vapnik-Chervonenkis dimension and the Rademacher complexity. An alternative approach, mainly studied in the statistical physics literature, is the study of generalization in simple synthetic-data models. Here we discuss the connections between these approaches and focus on the link between the Rademacher complexity in statistical learning and the theories of generalization for typical-case synthetic models from statistical physics, involving quantities known as Gardner capacity and ground state energy. We show that in these models the Rademacher complexity is closely related to the ground state energy computed by replica theories. Using this connection, one may reinterpret many results of the literature as rigorous Rademacher bounds in a variety of models in the high-dimensional statistics limit. Somewhat surprisingly, we also show that statistical learning theory provides predictions for the behavior of the ground-state energies in some full replica symmetry breaking models.

Tensor Recovery from Noisy and Multi-Level Quantized Measurements Ren Wang, Meng Wang, Jinjun Xiong Abstract --Higher-order tensors can represent scores in a rating system, frames in a video, and images of the same subject. In practice, the measurements are often highly quantized due to the sampling strategies or the quality of devices. Existing works on tensor recovery have focused on data losses and random noises. Only a few works consider tensor recovery from quantized measurements but are restricted to binary measurements. This paper, for the first time, addresses the problem of tensor recovery from multilevel quantized measurements. Leveraging the low-rank property of the tensor, this paper proposes a nonconvex optimization problem for tensor recovery. We provide a theoretical upper bound of the recovery error, which diminishes to zero when the sizes of dimensions increase to infinity. Our error bound significantly improves over the existing results in one-bit tensor recovery and quantized matrix recovery. A tensor-based alternating proximal gradient descent algorithm with a convergence guarantee is proposed to solve the nonconvex problem. Our recovery method can handle data losses and do not need the information of the quantization rule. The method is validated on synthetic data, image datasets, and music recommender datasets. I NTRODUCTION Many practical datasets are highly noisy and quantized, and recovering the actual values from quantized measurements finds applications in different domains.