There is no questioning that we as a species have entered a new age -- the age of information, or big data. Human beings are creating and consuming more and more data on an annual basis. Data that is not only becoming increasingly valuable but increasingly more energy-intensive. This problem is being addressed to some degree by switching to renewable energy sources like wind and solar but to truly tackle the issue, consumption needs to be reduced through technological innovation. Blockchain networks, in particular, have drawn criticism recently for their inefficient design and excessively high power consumption.

From biodegradable plastics to humanoid robots, a new wave of emerging technologies is on the horizon that have the potential to provide major benefits to societies and economies in the years to come, a new World Economic Forum (WEF) report said. An international Steering Committee of leading technology experts identified this year's "Top 10 Emerging Technologies" -- humanoid (and animaloid) robots designed to socialize with people; a system for pinpointing the source of a food-poisoning outbreak in seconds and minuscule lenses that will pave the way for diminutive cameras and other devices, among others. "Technologies that are emerging today will soon be shaping the world tomorrow and well into the future – with impacts to economies and to society at large," said Mariette DiChristina, Editor-in-Chief of Scientific American, and chair of the Emerging Technologies Steering Committee. Bioplastics are advanced solvents and enzymes that are transforming woody wastes into better biodegradable plastics. Like standard plastics derived from petrochemicals, biodegradable versions consist of polymers (long-chain molecules) that can be moulded while in their fluid state into a variety of forms.

Researchers launch and retrieve drones from a boat to photograph humpback and minke whales in the Western Antarctic Peninsula. BEAUFORT -- The swift pace of technological development has given researchers tools that can collect more data in less time and with fewer resources than a decade ago. Lightweight tags with long-lasting batteries can track animals as small as insects and measure the conditions around them. DNA sequencing technologies have decoded the genomes of thousands of organisms from the loblolly pine to the black bear. Drones can quickly photograph landscapes and animals in locations that may be inaccessible or unsafe for people.

You can think of artificial intelligence (AI) in the same way you think about cloud computing, if you think about either of them through an environmental lens: an enormous and growing source of carbon emissions, with the very real potential to choke out humans' ability to breathe clean air long before a sentient and ornery AI goes all Skynet on us. At the moment, data centers--the enormous rooms full of stacks and stacks of servers that juggle dank memes, fire tweets, your vitally important Google docs and all the other data that is stored somewhere other than on your phone and in your home computer--use about 2% of the world's electricity. Of that, servers that run AI--processing all the data and making the decisions and computations that a machine mimicking a human brain must handle in order to achieve "deep learning"--use about 0.1% of the world's electricity, according to a recent MIT Technology Review article. The likelihood that figure will grow, it turns out, is quite good. Until recently, more than a few scholarly dives into AI and the electricity grid focused on how AI could make power usage or other current carbon-emission-generating pastimes smarter.

In my earlier post, I wrote about the events leading up to my paper in GIScience & Remote Sensing. In this short post I would like to help you conduct your own machine learning classification of Sentinel-2 data using the open source package R. The process is pretty straightforward if you have experience in remote sensing and image classification. Even if you don't have extensive experience, basic knowledge of remote sensing terminology is sufficient. I've provided detailed information about different machine learning algorithms, including explanations of key concepts in my article linked below.

Over the past few years, brands have been toying with different ways AI can double as a wordsmith. While these experiments have proven that AI's ability to "learn" mass amounts of information give it a unique advantage when it comes to churning out copy, it's also become increasingly clear that there's only so much the technology can provide from a creative perspective. Take Saatchi & Saatchi Los Angeles, which trained IBM Watson to spit out copy for Toyota Mirai ads in 2017 as part of a campaign geared toward tech and science enthusiasts. While Watson was eventually able to unearth some interesting insights and string together clever lines of copy, getting there was a laborious process that involved months of training. Last year, Alibaba's digital marketing arm unveiled an AI-powered copywriting tool for brands to leverage on its ecommerce sites.

While the backpropagation of error algorithm allowed for a rapid rise in the development and deployment of artificial neural networks, two key issues currently preclude biological plausibility: (i) symmetry is required between forward and backward weights, which is known as the weight transport problem, and (ii) updates are locked before both the forward and backward passes have been completed. The feedback alignment (FA) algorithm uses fixed random feedback weights to release the weight transport problem. The direct feedback alignment (DFA) variation directly propagates the output error to each hidden layer through fixed random connectivity matrices. In this work, we show that using only the error sign is sufficient to maintain feedback alignment and to provide learning in the hidden layers. As in classification problems the error sign information is already contained in the target vector, using the latter as a proxy for the error brings three advantages: (i) it solves the weight transport problem by eliminating the requirement for an explicit feedback pathway, which also reduces the computational workload, (ii) it reduces memory requirements by removing update locking, allowing for weight updates to be computed in each layer independently without requiring a full forward pass, and (iii) it leads to a purely feedforward and low-cost algorithm that only requires a label-dependent random vector selection to estimate the layerwise loss gradients. Therefore, in this work, we propose the direct random target projection (DRTP) algorithm and demonstrate on the MNIST and CIFAR-10 datasets that, despite the absence of an explicit error feedback, DRTP performance can still lie close to the one of BP, FA and DFA. The low memory and computational cost of DRTP and its reliance only on layerwise feedforward computation make it suitable for deployment in adaptive edge computing devices.

Machine learning at the edge offers great benefits such as increased privacy and security, low latency, and more autonomy. However, a major challenge is that many devices, in particular edge devices, have very limited memory, weak processors, and scarce energy supply. We propose a hybrid hardware-software framework that has the potential to significantly reduce the computational complexity and memory requirements of on-device machine learning. In the first step, inspired by compressive sensing, data is collected in compressed form simultaneously with the sensing process. Thus this compression happens already at the hardware level during data acquisition. But unlike in compressive sensing, this compression is achieved via a projection operator that is specifically tailored to the desired machine learning task. The second step consists of a specially designed and trained deep network. As concrete example we consider the task of image classification, although the proposed framework is more widely applicable. An additional benefit of our approach is that it can be easily combined with existing on-device techniques. Numerical simulations illustrate the viability of our method.

Many real-world sequential decision-making problems can be formulated as optimal control with high-dimensional observations and unknown dynamics. A promising approach is to embed the high-dimensional observations into a lower-dimensional latent representation space, estimate the latent dynamics model, then utilize this model for control in the latent space. An important open question is how to learn a representation that is amenable to existing control algorithms? In this paper, we focus on learning representations for locally-linear control algorithms, such as iterative LQR (iLQR). By formulating and analyzing the representation learning problem from an optimal control perspective, we establish three underlying principles that the learned representation should comprise: 1) accurate prediction in the observation space, 2) consistency between latent and observation space dynamics, and 3) low curvature in the latent space transitions. These principles naturally correspond to a loss function that consists of three terms: prediction, consistency, and curvature (PCC). Crucially, to make PCC tractable, we derive an amortized variational bound for the PCC loss function. Extensive experiments on benchmark domains demonstrate that the new variational-PCC learning algorithm benefits from significantly more stable and reproducible training, and leads to superior control performance. Further ablation studies give support to the importance of all three PCC components for learning a good latent space for control.

Generative models have achieved impressive results in many domains including image and text generation. In the natural sciences, generative models have led to rapid progress in automated drug discovery. Many of the current methods focus on either 1-D or 2-D representations of typically small, drug-like molecules. However, many molecules require 3-D descriptors and exceed the chemical complexity of commonly used dataset. We present a method to encode and decode the position of atoms in 3-D molecules from a dataset of nearly 50,000 stable crystal unit cells that vary from containing 1 to over 100 atoms. We construct a smooth and continuous 3-D density representation of each crystal based on the positions of different atoms. Two different neural networks were trained on a dataset of over 120,000 three-dimensional samples of single and repeating crystal structures, made by rotating the single unit cells. The first, an Encoder-Decoder pair, constructs a compressed latent space representation of each molecule and then decodes this description into an accurate reconstruction of the input. The second network segments the resulting output into atoms and assigns each atom an atomic number. By generating compressed, continuous latent spaces representations of molecules we are able to decode random samples, interpolate between two molecules, and alter known molecules.