Distribution mismatches between the data seen at training and at application time remain a major challenge in all application areas of machine learning. We study this problem in the context of machine listening (Task 1b of the DCASE 2019 Challenge). We propose a novel approach to learn domain-invariant classifiers in an end-to-end fashion by enforcing equal hidden layer representations for domain-parallel samples, i.e. time-aligned recordings from different recording devices. No classification labels are needed for our domain adaptation (DA) method, which makes the data collection process cheaper.

We provide the construction of a set of square matrices whose translates and rotates provide a Parseval frame that is optimal for approximating a given dataset of images. Our approach is based on abstract harmonic analysis techniques. Optimality is considered with respect to the quadratic error of approximation of the images in the dataset with their projection onto a linear subspace that is invariant under translations and rotations. In addition, we provide an elementary and fully self-contained proof of optimality, and the numerical results from datasets of natural images.

Recent developments within deep learning are relevant for nonlinear system identification problems. In this paper, we establish connections between the deep learning and the system identification communities. It has recently been shown that convolutional architectures are at least as capable as recurrent architectures when it comes to sequence modeling tasks. Inspired by these results we explore the explicit relationships between the recently proposed temporal convolutional network (TCN) and two classic system identification model structures; Volterra series and block-oriented models. We end the paper with an experimental study where we provide results on two real-world problems, the well-known Silverbox dataset and a newer dataset originating from ground vibration experiments on an F-16 fighter aircraft.

While deep learning models become more widespread, their ability to handle unseen data and generalize for any scenario is yet to be challenged. In medical imaging, there is a high heterogeneity of distributions among images based on the equipment that generate them and their parametrization. This heterogeneity triggers a common issue in machine learning called domain shift, which represents the difference between the training data distribution and the distribution of where a model is employed. A high domain shift tends to implicate in a poor performance from models. In this work, we evaluate the extent of domain shift on three of the largest datasets of chest radiographs. We show how training and testing with different datasets (e.g. training in ChestX-ray14 and testing in CheXpert) drastically affects model performance, posing a big question over the reliability of deep learning models.

Artificial intelligence is one of a series of technologies that are on the radar for implementation in Chile's mining industry. AI, which, simply put, is the ability of a computer program or machine to think and learn from observing large quantities of data, to identify trends and make recommendations to improve decision making, all in a matter of milliseconds. The impending tsunami of data that will be collected from sensors and internet of things (IoT) devices will be too overwhelming for humans to compute. Businesses that are able to compute and extract value from huge volumes of data are expected to have a key advantage over their competitors by being able to improve efficiency, productivity and lower costs as well as identify new business opportunities. A recent study by consultancy Accenture, showed 82% of executives in the global mining industry expecting to increase investment in digital technology over the next three years.

Artificial Intelligence, Robotics, Machine learning-led technological innovation has already laid the foundation for higher productivity, better-income jobs, and socio-economic prosperity. In the coming years, automation will completely transform the nature and future of work, making things better and faster. However, these developments have also created the fear that the fourth industrial revolution or automation will lead to widespread labor displacement, lower wage growth, and worsen income inequality, especially in developing economies. The concerns may somehow be true as the automated technologies will replace aging and unskilled workforce with the new and technically skilled. As the Organisation for Economic Co-operation and Development has stated in its'OECD Employment Outlook 2019' report, "the risk of job automation is real but the trend varies greatly across countries. Automation technologies do not just destroy jobs, they also create and transform them. Historically, the net effects of major technological revolutions on employment have been positive, and there are few signs of this trend changing radically in the years to come."

A team of researchers from Bioversity International in Africa has created a smartphone app to help banana farmers protect their crops against diseases and pests. The Tumaini App (meaning'hope' in Swahili) is based on artificial intelligence algorithms that have been trained to recognize five major diseases and one common pest affecting the world's favorite fruit, demonstrating accuracy of more than 90 per cent in most models. The software has been tested in Colombia, the Democratic Republic of the Congo, India, Benin, China, and Uganda. Tumaini can recommend the means of addressing a specific disease and automatically upload identification data into a global database to help coordinate international response. It is hoped that the app can stop disease outbreaks and protect the livelihood of small, independent farmers.

This month, images of the burning Amazon rainforest have reverberated around the world. These fires, believed to have been set deliberately by cattle ranchers and loggers, have now spun out of control, leading to unprecedented destruction and dire warnings from environmentalists that the crisis will lead to the loss of a precious ecosystem and an acceleration of climate change. Brazil has rejected aid and the crisis has been blamed on the country's president Jair Bolsonaro, who critics say has encouraged farmers and loggers to burn far more of the forest than they typically do to clear land in order to graze cattle. But elsewhere, technology is helping some communities prevent and tackle wildfires...

When the data is distributed across multiple servers, efficient data exchange between the servers (or workers) for solving the distributed learning problem is an important problem and is the focus of this paper. We propose a fast, privacy-aware, and communication-efficient decentralized framework to solve the distributed machine learning (DML) problem. The proposed algorithm, GADMM, is based on Alternating Direct Method of Multiplier (ADMM) algorithm. The key novelty in GADMM is that each worker exchanges the locally trained model only with two neighboring workers, thereby training a global model with lower amount of communication in each exchange. We prove that GADMM converges faster than the centralized batch gradient descent for convex loss functions, and numerically show that it is faster and more communication-efficient than the state-of-the-art communication-efficient centralized algorithms such as the Lazily Aggregated Gradient (LAG), in linear and logistic regression tasks on synthetic and real datasets. Furthermore, we propose Dynamic GADMM (D-GADMM), a variant of GADMM, and prove its convergence under time-varying network topology of the workers.

India's higher education sector has supplied some of the world's best talent. The CEOs of some of the biggest Fortune 500 companies--Microsoft, Google, Mastercard, and Adobe--are a product of the Indian higher education system. The landscape has also expanded over the past decade--from 436 universities in 2009–10 to 903 in 2017–18 and from 26,000 colleges to over 39,000.1 Student enrolment, at 36.6 million, is the third-largest in the world, next to China and the United States.2 Besides, India is already in the middle of the "demographic dividend" with a surge in its younger and working-age population, which is estimated to become the world's largest by 2030.3 India is expected to account for about 20 percent of the total young talent pool supplied by the non–Organisation for Economic Cooperation and Development (OECD) G-20 countries.4