Raytheon Technologies (Finland)

In silico modeling is a crucial milestone in modern drug design and development. Although computer-aided approaches in this field are well-studied, the application of deep learning methods in this research area is at the beginning. In this work, we present an original deep neural network (DNN) architecture named RANC (Reinforced Adversarial Neural Computer) for the de novo design of novel small-molecule organic structures based on the generative adversarial network (GAN) paradigm and reinforcement learning (RL). As a generator RANC uses a differentiable neural computer (DNC), a category of neural networks, with increased generation capabilities due to the addition of an explicit memory bank, which can mitigate common problems found in adversarial settings. The comparative results have shown that RANC trained on the SMILES string representation of the molecules outperforms its first DNN-based counterpart ORGANIC by several metrics relevant to drug discovery: the number of unique structures, passing medicinal chemistry filters (MCFs), Muegge criteria, and high QED scores. RANC is able to generate structures that match the distributions of the key chemical features/descriptors (e.g., MW, logP, TPSA) and lengths of the SMILES strings in the training data set. Therefore, RANC can be reasonably regarded as a promising starting point to develop novel molecules with activity against different biological targets or pathways. In addition, this approach allows scientists to save time and covers a broad chemical space populated with novel and diverse compounds.

Current climate control systems often rely on building regulation maximum occupancy numbers for maintaining proper temperatures. However, in many situations, there are rooms that are used infrequently, and may be heated or cooled needlessly. Having knowledge regarding occupancy and being able to accurately predict usage patterns may al-low significant energy-savings by intelligent control of the L-HVAC systems. In this paper, we report on the deploy-ment of a wireless camera sensor network for collecting data regarding occupancy in a large multi-function building. The system estimates occupancy with an accuracy of 80%. Using data collected from this system, we construct multivariate Gaussian and agent based models for predicting user mobil-ity patterns in buildings. Using these models, we can predict room usage thereby enabling us to control the HVAC systems in an adaptive manner. Our simulations indicate a 14 % re-duction in HVAC energy usage by having an optimal control strategy based on occupancy estimates and usage patterns. 1

-rich fragments, and IDC). The SMILES representations of 15K druglike molecules from the ChemDiv collection were used as a training data set. For the different functions, ATNC outperforms ORGANIC. Combined with the IDC, ATNC generates 72% of valid and 77% of unique SMILES strings, while ORGANIC generates only 7% of valid and 86% of unique SMILES strings. For each set of molecules generated by ATNC and ORGANIC, we analyzed distributions of four molecular descriptors (number of atoms, molecular weight, logP, and tpsa) and calculated five chemical statistical features (internal diversity, number of unique heterocycles, number of clusters, number of singletons, and number of compounds that have not been passed through medicinal chemistry filters). Analysis of key molecular descriptors and chemical statistical features demonstrated that the molecules generated by ATNC elicited better druglikeness properties. We also performed in vitro validation of the molecules generated by ATNC; results indicated that ATNC is an effective method for producing hit compounds.

Liquefied gas electrolytes with unique solvation structure enable high ionic conductivity in extended temperature ranges, supporting wide-temperature high-voltage lithium metal batteries.

A new nonaqueous symmetric redox flow battery was developed based on an organic ambipolar electroactive material. FTIR demonstrated great potential for online monitoring of the state of charge of this flow battery.

Local rates of heat transfer on the endwall, suction, and pressure surfaces of a large scale turbine blade cascade were measured for two inlet boundary layer thicknesses and for a Reynolds number typical of gas turbine engine operation. The accuracy and spatial resolution of the measurements were sufficient to reveal local variations of heat transfer associated with distinct flow regimes and with regions of strong three-dimensional flow. Pertinent results of surface flow visualization and pressure measurements are included. The dominant role of the passage vortex, which develops from the singular separation of the inlet boundary layer, in determining heat transfer at the endwall and at certain regions of the airfoil surface is illustrated. Heat transfer on the passage surfaces is discussed and measurements at airfoil midspan are compared with current finite difference prediction methods.

The purpose of this study was to investigate seasonal variations in population monthly hemoglobin A(1c) (A1c) values over 2 years (from October 1998 to September 2000) among US diabetic veterans. The study cohort included 285,705 veterans with 856,181 A1c tests. The authors calculated the monthly average A1c values for the overall population and for subpopulations defined by age, sex, race, insulin use, and climate regions. A1c values were higher in winter and lower in summer with a difference of 0.22. The proportion of A1c values greater than 9.0% followed a similar seasonal pattern that varied from 17.3% to 25.3%. Seasonal autoregressive models including trigonometric function terms were fit to the monthly average A1c values. There were significant seasonal effects; the seasonal variation was consistent across different subpopulations. Regions with colder winter temperatures had larger winter-summer contrasts than did those with warmer winter temperatures. The seasonal patterns followed trends similar to those of many physiologic markers, cardiovascular and other diabetes outcomes, and mortality. These findings have implications for health-care service research in quality-of-care assessment, epidemiologic studies investigating population trends and risk factors, and clinical trials or program evaluations of treatments or interventions.

This chapter contains sections titled: Introduction Electronic Structure Calculations Electronic Structure and Potential-Energy Curves

In recent years, Deep Learning has been successfully applied to multimodal learning problems, with the aim of learning useful joint representations in data fusion applications. When the available modalities consist of time series data such as video, audio and sensor signals, it becomes imperative to consider their temporal structure during the fusion process. In this paper, we propose the Correlational Recurrent Neural Network (CorrRNN), a novel temporal fusion model for fusing multiple input modalities that are inherently temporal in nature. Key features of our proposed model include: (i) simultaneous learning of the joint representation and temporal dependencies between modalities, (ii) use of multiple loss terms in the objective function, including a maximum correlation loss term to enhance learning of cross-modal information, and (iii) the use of an attention model to dynamically adjust the contribution of different input modalities to the joint representation. We validate our model via experimentation on two different tasks: video-and sensor-based activity classification, and audio-visual speech recognition. We empirically analyze the contributions of different components of the proposed CorrRNN model, and demonstrate its robustness, effectiveness and state-of-the-art performance on multiple datasets.

Rogue (unauthorized) wireless access points pose serious security threats to local networks. In this paper, we propose two online algorithms to detect rogue access points using sequential hypothesis tests applied to packet-header data collected passively at a monitoring point. One algorithm requires training sets, while the other does not. Both algorithms extend our earlier TCP ACK-pair technique to differentiate wired and wireless LAN TCP traffic, and exploit the fundamental properties of the 802.11 CSMA/CA MAC protocol and the half duplex nature of wireless channels. Our algorithms make prompt decisions as TCP ACK-pairs are observed, and only incur minimum computation and storage overhead. We have built a system for online rogue-access-point detection using these algorithms and deployed it at a university gateway router. Extensive experiments in various scenarios have demonstrated the excellent performance of our approach: the algorithm that requires training provides rapid detection and is extremely accurate (the detection is mostly within 10 seconds, with very low false positive and false negative ratios); the algorithm that does not require training detects 60%-76% of the wireless hosts without any false positives; both algorithms are light-weight (with computation and storage overhead well within the capability of commodity equipment).

A major challenge in designing cancer therapies is the induction of cancer cell apoptosis, although activation of intrinsic apoptotic pathways by targeting gold nanoparticles to mitochondria is promising. We report an in vitro procedure targeting mitochondria with conjugated gold nanoparticles and investigating effects on apoptosis induction in the human breast cancer cell line Jimt-1. Gold nanoparticles were conjugated to a variant of turbo green fluorescent protein (mitoTGFP) harbouring an amino-terminal mitochondrial localization signal. Au nanoparticle conjugates were further complexed with cationic maltotriose-modified poly(propylene imine) third generation dendrimers. Fluorescence and transmission electron microscopy revealed that Au nanoparticle conjugates were directed to mitochondria upon transfection, causing partial rupture of the outer mitochondrial membrane, triggering cell death. The ability to target Au nanoparticles into mitochondria of breast cancer cells and induce apoptosis reveals an alternative application of Au nanoparticles in photothermal therapy of cancer.

A trade-factor-based system study has been carried out to identify fuel burn benefits associated with boundary layer ingestion (BLI) for generation-after-next (N+2) aircraft and propulsion system concepts. The analysis includes detailed propulsion system engine cycle modeling for a next-generation, Ultra-High-Bypass (UHB) propulsion system with BLI using the Numerical Propulsion System Simulation (NPSS) computational model. Cycle modeling was supplemented with one-dimensional theory to identify limiting theoretical BLI benefits associated with the blended wing body reference vehicle used in the study. The system study employed low-order models of engine extractions associated with inlet flow control; nacelle weight and drag; fan performance; and inlet pressure losses. Aircraft trade factors were used to estimate block fuel burn reduction for a long-range commercial transport mission. Results of the study showed that a 3-5% BLI fuel burn benefit can be achieved for N+2 aircraft relative to a baseline high-performance, pylon-mounted, UHB propulsion system. High-performance, distortion-tolerant turbomachinery, and low-loss, low-drag inlet systems, were identified as key enabling technologies. Larger benefits were estimated for N+3 configurations for which larger fractions of aircraft boundary layer can be ingested.

Redox flow batteries possess several key advantages that make them well suited for grid-scale energy-storage applications. However, the capital cost of flow batteries has been a major barrier to commercialization of this technology. One attractive path to cost reduction is the development of flow-battery cells with substantially higher power densities than conventional flow-battery cells. The cost of the cells comprises a significant portion of the total flow-battery system cost, especially at low production volumes, since cell parts are custom-built components. UTRC has developed high power density vanadium-redox battery cells utilizing a material set similar to conventional cells. This advanced cell technology can theoretically be applied to other flow-battery chemistries as well. This breakthrough in cell performance should motivate additional development of flow-battery technologies, since a realistic path to substantial cost reduction, which can be realized even at low production volumes, has now been demonstrated.

N-containing heterocycles can tune the electrochemical properties of organodisulfides in lithium batteries. 2,2′-Dipyridyl disulfide exhibits outstanding cycling stability.

Membrane failure results from a complex interplay of mechanical stress and chemical attack; each have multiple origins. A picture of degradation is emerging where the polymer is attacked by hydroxyl-radicals generated on Pt particles within the membrane. We extend this picture to include several reactions of hydroxyl radicals with peroxide and the PFSA polymer. This extended picture allows us to explain three differences in behavior exhibited by membrane-degradation rates in Fenton's Reagent tests and in-cell tests regarding correlation of rate with end groups, spatial distribution of polymer loss and existence of induction periods. We postulate a 3-step degradation scenario: 1) hydroxyl radicals attack polymer end groups and side chains, 2) hydroxyl radicals are converted to peroxyl radicals which only attack polymer end groups, and 3) loss of polymer volume and membrane embrittlement leading to mechanical failure depending on the extent of degradation.

The dynamics of bluff body stabilized flames are investigated using an unsteady, Lagrangian simulation technique . This paper presents an analysis of results from a parametric study of temperature ratio across a flame stabilized on a bluff body confined in a duct . The simulation presented in this study is a two -dimensional description of a premixed inflow of reactants with flame sheets attached to a triangular wedge. The analysis concentrates on flow field dynamics related to the transition from the asym metric Von Karman shedding found in non -reacting bluff body flows to independent shear layer vorticity commonly observed in many reacting flows. Details of this transition are described in terms of the instantaneous vorticity field, as well as mean and uns teady components of velocity and temperature. It is shown that for relatively small temperature differences across the flamesheets, the Von Karman shedding mode remains dominant, although the strength of the shed vortices decreases with increasing temperat ure ratio. As the temperature ratio increases, the Von Karman shedding mode is completely suppressed, and this suppression is clearly evident through RMS temperature profiles as well as turbulent stresses near the bluff body.

This article continues an investigation begun in [2]. A random graph G n ( x ) is constructed on independent random points U 1 , · ··, U n distributed uniformly on [0, 1] d , d ≧ 1, in which two distinct such points are joined by an edge if the l ∞ -distance between them is at most some prescribed value 0 < x < 1. Almost-sure asymptotic results are obtained for the convergence/divergence of the minimum vertex degree of the random graph, as the number n of points becomes large and the edge distance x is allowed to vary with n. The largest nearest neighbor link d n , the smallest x such that G n ( x ) has no vertices of degree zero, is shown to satisfy Series and sequence criteria on edge distances {x n } are provided which guarantee the random graph to be complete, a.s. These criteria imply a.s. limiting behavior of the diameter of the vertex set.

This article focuses on the design of systems in which a human operator is responsible for overseeing autonomous agents and providing feedback based on sensor data. In the control systems community, the term human supervisory control (or simply supervisory control) is often used as a shorthand reference for systems with this type of architecture [5]-[7]. In a typical human supervisory control application, the operator does not directly manipulate autonomous agents but rather indirectly interacts with these components via a central data-processing station. As such, system designers have the opportunity to easily incorporate automated functionalities to control how information is presented to the operator and how the input provided by the operator is used by automated systems. The goal of these functionalities is to take advantage of the inherent robustness and adaptability of human operators, while mitigating adverse effects such as unpredictability and performance variability. In some contexts, to meet the goal of single-operator supervision of multiple automated sensor systems, such facilitating mechanisms are not only useful but necessary for practical use [8], [9]. A successful system design must carefully consider the goals of each part of the system as a whole and seamlessly stitch components together using facilitating functionalities.

High speed exhaust noise reduction continues to be a research challenge for supersonic cruise business jets as well as for current and future tactical military aircraft. Significant noise reduction may be possible from advanced concepts for controlling instability generated large-scale turbulence structures in the jet shear layer, generally accepted to be the source of aft-angle noise. In response to this opportunity, our team is focused on experimental diagnostic studies and unique instability modeling suited for identifying control strategies to reduce large scale structure noise. The current paper benchmarks the jet noise from supersonic nozzles designed to provide the supporting experimental data and validation of the modeling. Laboratory scale jet noise experiments are presented for a Mach number of Mj = 1.5 with stagnation temperature ratios ranging from Tr=0.75 to 2. The baseline configuration is represented by a round converging-diverging (CD) ideal expansion nozzle. A round CD nozzle with chevrons is included as the first of several planned non-circular geometries directed at demonstrating the impact on large scale structure noise and validating noise prediction methods for geometries of future technological interest. Overexpanded and

Abstract Morphological changes in knee cartilage subregions are valuable imaging‐based biomarkers for understanding progression of osteoarthritis, and they are typically detected from magnetic resonance imaging (MRI). So far, accurate segmentation of cartilage has been done manually. Deep learning approaches show high promise in automating the task; however, they lack clinically relevant evaluation. We introduce a fully automatic method for segmentation and subregional assessment of articular cartilage, and evaluate its predictive power in context of radiographic osteoarthritis progression. Two data sets of 3D double‐echo steady‐state (DESS) MRI derived from the Osteoarthritis Initiative were used: first, n = 88; second, n = 600, 0‐/12‐/24‐month visits. Our method performed deep learning‐based segmentation of knee cartilage tissues, their subregional division via multi‐atlas registration, and extraction of subregional volume and thickness. The segmentation model was developed and assessed on the first data set. Subsequently, on the second data set, the morphological measurements from our and the prior methods were analyzed in correlation and agreement, and, eventually, by their discriminative power of radiographic osteoarthritis progression over 12 and 24 months, retrospectively. The segmentation model showed very high correlation ( r > 0.934) and agreement (mean difference < 116 mm 3 ) in volumetric measurements with the reference segmentations. Comparison of our and manual segmentation methods yielded r = 0.845–0.973 and mean differences = 262–501 mm 3 for weight‐bearing cartilage volume, and r = 0.770–0.962 and mean differences = 0.513–1.138 mm for subregional cartilage thickness. With regard to osteoarthritis progression, our method found most of the significant associations identified using the manual segmentation method, for both 12‐ and 24‐month subregional cartilage changes. The method may be effectively applied in osteoarthritis progression studies to extract cartilage‐related imaging biomarkers.