We use data on 124 batteries released by Stanford University to first try to solve the binary classification problem of determining if a battery is "good" or "bad" given only the first 5 cycles of data (i.e., will it last longer than a certain threshold of cycles), as well as the prediction problem of determining the exact number of cycles a battery will last given the first 100 cycles of data. We approach the problem from a purely data-driven standpoint, hoping to use deep learning to learn the patterns in the sequences of data that the Stanford team engineered by hand. For both problems, we used a similar deep network design, that included an optional 1-D convolution, LSTMs, an optional Attention layer, followed by fully connected layers to produce our output. For the classification task, we were able to achieve very competitive results, with validation accuracies above 90%, and a test accuracy of 95%, compared to the 97.5% test accuracy of the current leading model. For the prediction task, we were also able to achieve competitive results, with a test MAPE error of 12.5% as compared with a 9.1% MAPE error achieved by the current leading model (Severson et al. 2019).

But as these organizations grapple with growing demand, erratic temperatures, aging infrastructure, and the threat of cyberattacks, many struggle to maintain a high level of service in an uncertain and unpredictable landscape. Artificial intelligence (AI) and machine learning (ML), as powered by big data, have the potential to modernize energy and utilities organizations by identifying ways to reduce waste and redundancy, protect and manage assets, and detect performance anomalies – all while realizing valuable cost savings, both for the organization and the customer. In this blog, we explore the three main areas where AI is making a mark on the energy and utilities sector today and how such investments may impact the future. Each year in the U.S. alone, trillions of gallons of water are lost due to aging pipes, broken water mains, and faulty meters. Replacing the entire system would be massively expensive, time-consuming, and impractical, which means that utility companies must take a localized approach to repairs.

If we were to go back 50 years in time and demonstrate modern technology to someone, it might appear indistinguishable from magic. To have within the palm of your hand one device that you can use to send instant messages, read books, pay bills, make movies, and even find love would seem unimaginable. Yet this reality has all been made possible by key breakthrough technologies that were revolutionary -- and, some might argue, inevitable. Kevin Kelly, co-founder of Wired magazine, once observed that grand-scale technologies are predictable because they have an inherent direction. He uses the analogy of gravity.

Perhaps no technology has generated more hype in recent years than artificial intelligence (AI). In some industries, it is certainly living up to it.

Pacific Northwest National Laboratory is joining forces with two other research powerhouses to pioneer a new $5.5 million research center created by the U.S. Department of Energy to focus on the biggest challenges in artificial intelligence. The Center for Artificial Intelligence-Focused Architectures and Algorithms, or ARIAA, will promote collaborative projects for scientists at PNNL in Richland, Wash., at Sandia National Laboratories in New Mexico, and at Georgia Tech. PNNL and Sandia are part of the Energy Department's network of research labs. ARIAA will be headed by Roberto Gioiosa, a senior research scientist at PNNL. As center director, he'll be in charge of ARIAA's overall vision, strategy and research direction.

In point-based sensing systems such as coordinate measuring machines (CMM) and laser ultrasonics where complete sensing is impractical due to the high sensing time and cost, adaptive sensing through a systematic exploration is vital for online inspection and anomaly quantification. Most of the existing sequential sampling methodologies focus on reducing the overall fitting error for the entire sampling space. However, in many anomaly quantification applications, the main goal is to estimate sparse anomalous regions in the pixel-level accurately. In this paper, we develop a novel framework named Adaptive Kernelized Maximum-Minimum Distance AKM$^2$D to speed up the inspection and anomaly detection process through an intelligent sequential sampling scheme integrated with fast estimation and detection. The proposed method balances the sampling efforts between the space-filling sampling (exploration) and focused sampling near the anomalous region (exploitation). The proposed methodology is validated by conducting simulations and a case study of anomaly detection in composite sheets using a guided wave test.

The use of Reinforcement Learning (RL) is still restricted to simulation or to enhance human-operated systems through recommendations. Real-world environments (e.g. industrial robots or power grids) are generally designed with safety constraints in mind implemented in the shape of valid actions masks or contingency controllers. For example, the range of motion and the angles of the motors of a robot can be limited to physical boundaries. Violating constraints thus results in rejected actions or entering in a safe mode driven by an external controller, making RL agents incapable of learning from their mistakes. In this paper, we propose a simple modification of a state-of-the-art deep RL algorithm (DQN), enabling learning from forbidden actions. To do so, the standard Q-learning update is enhanced with an extra safety loss inspired by structured classification. We empirically show that it reduces the number of hit constraints during the learning phase and accelerates convergence to near-optimal policies compared to using standard DQN. Experiments are done on a Visual Grid World Environment and Text-World domain.

This work extends the randomized shortest paths model (RSP) by investigating the net flow RSP and adding capacity constraints on edge flows. The standard RSP is a model of movement, or spread, through a network interpolating between a random walk and a shortest path behavior. This framework assumes a unit flow injected into a source node and collected from a target node with flows minimizing the expected transportation cost together with a relative entropy regularization term. In this context, the present work first develops the net flow RSP model considering that edge flows in opposite directions neutralize each other (as in electrical networks) and proposes an algorithm for computing the expected routing costs between all pairs of nodes. This quantity is called the net flow RSP dissimilarity measure between nodes. Experimental comparisons on node clustering tasks show that the net flow RSP dissimilarity is competitive with other state-of-the-art techniques. In the second part of the paper, it is shown how to introduce capacity constraints on edge flows and a procedure solving this constrained problem by using Lagrangian duality is developed. These two extensions improve significantly the scope of applications of the RSP framework.

Recent work has shown that quantization can help reduce the memory, compute, and energy demands of deep neural networks without significantly harming their quality. However, whether these prior techniques, applied traditionally to image-based models, work with the same efficacy to the sequential decision making process in reinforcement learning remains an unanswered question. To address this void, we conduct the first comprehensive empirical study that quantifies the effects of quantization on various deep reinforcement learning policies with the intent to reduce their computational resource demands. We apply techniques such as post-training quantization and quantization aware training to a spectrum of reinforcement learning tasks (such as Pong, Breakout, BeamRider and more) and training algorithms (such as PPO, A2C, DDPG, and DQN). Across this spectrum of tasks and learning algorithms, we show that policies can be quantized to 6-8 bits of precision without loss of accuracy. We also show that certain tasks and reinforcement learning algorithms yield policies that are more difficult to quantize due to their effect of widening the models' distribution of weights and that quantization aware training consistently improves results over post-training quantization and oftentimes even over the full precision baseline. Finally, we demonstrate real-world applications of quantization for reinforcement learning. We use half-precision training to train a Pong model 50% faster, and we deploy a quantized reinforcement learning based navigation policy to an embedded system, achieving an 18$\times$ speedup and a 4$\times$ reduction in memory usage over an unquantized policy.

If your home is in a wildfire area, insurance companies tend to not want to go anywhere near it. But "wildfire areas" tend to be pretty broad. What if companies could evaluate the risk on a more granular level -- tapping things like satellite imagery and machine learning combined with wind, weather and topology data, to better define the riskiest zones? Could more home owners be offered policies, and at more affordable rates? Delos itself doesn't act as the insurer; instead, it acts as a Managing General Agent (or MGA) for a bunch of major carriers.