We present a novel variant of Domain Adversarial Networks with impactful improvements to the loss functions, training paradigm, and hyperparameter optimization. New loss functions are defined for both forks of the DANN network, the label predictor and domain classifier, in order to facilitate more rapid gradient descent, provide more seamless integration into modern neural networking frameworks, and allow previously unavailable inferences into network behavior. Using these loss functions, it is possible to extend the concept of 'domain' to include arbitrary user defined labels applicable to subsets of the training data, the test data, or both. As such, the network can be operated in either 'On the Fly' mode where features provided by the feature extractor indicative of differences between 'domain' labels in the training data are removed or in 'Test Collection Informed' mode where features indicative of difference between 'domain' labels in the combined training and test data are removed (without needing to know or provide test activity labels to the network). This work also draws heavily from previous works on Robust Training which draws training examples from a L_inf ball around the training data in order to remove fragile features induced by random fluctuations in the data. On these networks we explore the process of hyperparameter optimization for both the domain adversarial and robust hyperparameters. Finally, this network is applied to the construction of a binary classifier used to identify the presence of EM signal emitted by a turbopump. For this example, the effect of the robust and domain adversarial training is to remove features indicative of the difference in background between instances of operation of the device - providing highly discriminative features on which to construct the classifier.

We present an approach to data fusion that combines the interpretability of structured probabilistic graphical models with the flexibility of neural networks. The proposed method, lightweight data fusion (LDF), emphasizes posterior analysis over latent variables using two types of information: primary data, which are well-characterized but with limited availability, and auxiliary data, readily available but lacking a well-characterized statistical relationship to the latent quantity of interest. The lack of a forward model for the auxiliary data precludes the use of standard data fusion approaches, while the inability to acquire latent variable observations severely limits direct application of most supervised learning methods. LDF addresses these issues by utilizing neural networks as conjugate mappings of the auxiliary data: nonlinear transformations into sufficient statistics with respect to the latent variables. This facilitates efficient inference by preserving the conjugacy properties of the primary data and leads to compact representations of the latent variable posterior distributions. We demonstrate the LDF methodology on two challenging inference problems: (1) learning electrification rates in Rwanda from satellite imagery, high-level grid infrastructure, and other sources; and (2) inferring county-level homicide rates in the USA by integrating socio-economic data using a mixture model of multiple conjugate mappings.

As part of the transportation authorities' efforts to address this problem, researchers from across the US Department of Energy's (DOE) Argonne National Laboratory in collaboration with the Lawrence Berkeley National Laboratory (LBNL) have developed a new artificial intelligence model to help alleviate congestion on the city's streets. That data was then used to train a model to forecast traffics, congestion spots, and average speed of cars on the routes. The new model can look at the past hour, and then predict the next hour of traffic with great accuracy within milliseconds. "The AI and supercomputing capabilities that have been used in this work allow us to tackle really large problems. The scale of this project is large, and this amount of data requires an equally large computing resource to tackle it," said Prasanna Balaprakash, a computer scientist in Argonne National Laboratory.

Energy, Oil and gas wealth is considered as one of the most valued commodities across the globe. As the revenue in the sector spikes, the security risks it faces also increases with more cyber and physical attacks taking place in the recent years. Unlike other industries, energy, oil & gas sector get a big bang on its whole working system every time a security breach happens. Ultimately, video analytics combined with artificial intelligence (AI) is shoving hackers the exit door with its extended technological influence. According to a report, global energy was valued at US$1.7 trillion, which is a 2.2% of the global GDP in 2016.

You can now run computations on your phone that would have been unthinkable a few years ago. But as small devices get smarter, we discover new uses for them that overwhelm their resources. If you want your phone to recognize a picture of your face (image classification) or to find faces in pictures (object detection), you want it to run a convolutional neural net (CNN). Modern computer vision applications are mostly built using CNNs. This is because vision applications tend to have a classifier at their heart--so, for example, one builds an object detector by building one classifier that tells whether locations in an image could contain an object, then another that determines what the object is.

Since 2012, the field of artificial intelligence (AI) has reported remarkable progress on a broad range of capabilities including object recognition, game playing, speech recognition, and machine translation.43 Much of this progress has been achieved by increasingly large and computationally intensive deep learning models.a Figure 1, reproduced from Amodei et al.,2 plots training cost increase over time for state-of-the-art deep learning models starting with AlexNet in 201224 to AlphaZero in 2017.45 The chart shows an overall increase of 300,000x, with training cost doubling every few months. An important paper47 has estimated the carbon footprint of several NLP models and argued this trend is both environmentally unfriendly and prohibitively expensive, raising barriers to participation in NLP research. We refer to such work as Red AI. The amount of compute used to train deep learning models has increased 300,000x in six years. Figure taken from Amodei et al.2 This trend is driven by the strong focus of the AI community on obtaining "state-of-the-art" results,b as exemplified by the popularity of leaderboards,53,54 which typically report accuracy (or other similar measures) but omit any mention of cost or efficiency (see, for example, leaderboards.allenai.org).c Despite the clear benefits of improving model accuracy, the focus on this single metric ignores the economic, environmental, and social cost of reaching the reported results.

The original version of this paper is entitled "SATURN: A thin and flexible self-powered microphone designed on the principle of triboelectric nanogenerator" and was published in Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies 2.2 (60), 2018, ACM

MIT CSAIL's flexible sensors can be applied as skin to the bodies of soft robots. When you picture a robot, you likely envision one large and rigid, with limited movement and an outer shell that is hard to the touch. Several projects currently underway seek to change that, with the use of soft, more human-like artificial skin. Artificial skins include any surface-based device or distributed network of sensors that enable an agent to perceive mechanical deformations, touch, temperature, vibration, and/or pain, according to Ryan Truby, a post-doctoral fellow in the Massachusetts Institute of Technology (MIT) Computer Science & Artificial Intelligence Lab (CSAIL). Engineers are working to create skins that include as many of these sensations as possible, while also possessing high sensitivity and spatial resolution in sensing, he adds.

The potential positive impact of Artificial Intelligence (AI) is significant and organizations can expect to cut GHG emissions by 16% in the next three to five years through AI-driven climate action projects according to a research report by Capgemini Research Institute. Despite the considerable potential of AI for climate action, adoption remains low. More than eight in ten organizations spend less than 5% of climate change investment on AI and data tracking; 54% have fewer than 5% of employees with the skills to take up data and AI-driven roles; and more than a third (37%) of sustainability executives have decelerated their climate goals in light of COVID-19, with the highest deceleration in the energy and utilities industry. Only 13% of organizations have aligned their climate vision and strategy with their AI capabilities – these are who Capgemini defines as climate AI champions. Two-fifths of these come from Europe, followed by the Americas and APAC.

The Dubai Electricity and Water Authority (Dewa) has adopted the use of Smart Dubai's Ethical AI Toolkit. It reports using it for 13 artificial intelligence (AI) use cases across various departments, registering an average performance rate of almost 90 per cent on complying with the principles and guidelines set out. Smart Dubai developed the toolkit to set clear guidelines on the ethical use of AI to prevent having a fragmented, incoherent approach to ethics, where every entity sets its own rules. Dewa's use of the toolkit was spread across several different departments. The Innovation & the Future (I&TF) division's use cases included outage planning and load forecasting, solar power generation forecasting, network design and area planning, visual inspection on solar photovoltaics and the virtual assistant Rammas.