Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Distance Learning”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 307 records · Page 17

Experimental High Energy Physics at the University of Illinois

This research effort funded by the Office of High Energy Physics in the U.S. Department of Energy is aimed at exploring our universe at its most basic level. The goal of our effort is to learn more about how and why nature behaves the way it does. Within this effort, we are studying the smallest particles and the largest distances we can possibly observe. Our work in particular focuses on the search for new interactions that will help us understand how the universe came into being. Through this work, we collaborate with other scientists and engineers to develop new technologies that can be utilized throughout society. Benefits from our research include advances in medical technology, electronics, transportation, sustainability, information and computing as well as the advanced training of undergraduate and graduate students in science and engineering.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Adaptive fuzzy leader clustering of complex data sets in pattern recognition

A modular, unsupervised neural network architecture for clustering and classification of complex data sets is presented. The adaptive fuzzy leader clustering (AFLC) architecture is a hybrid neural-fuzzy system that learns on-line in a stable and efficient manner. The initial classification is performed in two stages: a simple competitive stage and a distance metric comparison stage. The cluster prototypes are then incrementally updated by relocating the centroid positions from fuzzy C-means system equations for the centroids and the membership values. The AFLC algorithm is applied to the Anderson Iris data and laser-luminescent fingerprint image data. It is concluded that the AFLC algorithm successfully classifies features extracted from real data, discrete or continuous.

Newton, Scott C.↗

Deep Learning Classification of Cheatgrass Invasion in the Western United States Using Biophysical and Remote Sensing Data

Cheatgrass (Bromus tectorum) invasion is driving an emerging cycle of increased fire frequency and irreversible loss of wildlife habitat in the western US. Yet, detailed spatial information about its occurrence is still lacking for much of its presumably invaded range. Deep learning (DL) has demonstrated success for remote sensing applications but is less tested on more challenging tasks like identifying biological invasions using sub-pixel phenomena. We compare two DL architectures and the more conventional Random Forest and Logistic Regression methods to improve upon a previous effort to map cheatgrass occurrence at >2% canopy cover. High-dimensional sets of biophysical, MODIS, and Landsat-7 ETM+ predictor variables are also compared to evaluate different multi-modal data strategies. All model configurations improved results relative to the case study and accuracy generally improved by combining data from both sensors with biophysical data. Cheatgrass occurrence is mapped at 30 m ground sample distance (GSD) with an estimated 78.1% accuracy, compared to 250-m GSD and 71% map accuracy in the case study. Furthermore, DL is shown to be competitive with well-established machine learning methods in a limited data regime, suggesting it can be an effective tool for mapping biological invasions and more broadly for multi-modal remote sensing applications.

54 ENVIRONMENTAL SCIENCES↗

Improving the Accuracy of Clustering Electric Utility Net Load Data using Dynamic Time Warping

Identifying patterns in electric utility net load data in a time-series format is very useful in preparing the operation for next day. Machine learning algorithms have been used in other domains and those concepts are applied in this paper on real-world net load measurement data. Clustering is the practice of grouping data with similar characteristics as determined by the distance measure. The K-means clustering algorithm is utilized here with actual electric utility data. The paper uses the standard distance measure, Euclidean distance (ED), and compares its performance against the dynamic time warping (DTW) measure. An actual case study with real data is presented, and DTW distance measure-based method observed to result better accuracy compared to the ED based method for substation net load measurements predominantly with residential customers.

clustering↗

Characterizing the Spread of COVID-19 from Human Mobility Patterns and SocioDemographic Indicators

Mobility is an indicator of human movement through space and time. With the increasing availability of geolocated data (from GPS, accelerometers, etc.), it is now possible to examine individual as well as group human mobility patterns. Human mobility is influenced by both intrinsic (i.e. personal motivations) and extrinsic (i.e., events like natural hazards or a pandemic like the COVID-19) factors. However, the intricate relationships between human mobility patterns and sociodemographic characteristics in the context of a pandemic are yet to be fully explored. Our goal is to overcome this gap by using human mobility data at the census block group level from mobile phones and combining those with social vulnerability indicators to examine the overall spread of COVID-19 at local spatial scales. We used 585,878 weekly visits to 37,871 points of interests (POIs) from Safegraph to quantify mobility indices and social distancing metrics in 2,820 census block groups in the city of Los Angeles (LA) - before and during lockdown as well as during the phase1 and phase 2 reopening. Finally, using supervised machine learning algorithms, we classified the census block groups in LA into High, Medium and Low categories that represented the vulnerability of these block groups based on the cumulative number of occurrences of COVID-19 cases till July 24, 2020. Our results indicate that the tree-based classifiers performed well in comparison to the Support Vector Machines and Multinomial Logit models. Gradient Boosting had the highest classification accuracy of 97.4% COVID-19 with an AUC score of 0.987. The block groups with high COVID-19 cases also had a high concentration of socially vulnerable populations, high human mobility index and a low social distancing index.

Roy, Avipsa↗

Locally coupled open subsystems: A formalism for affordable electronic structure calculations featuring fractional charges and size consistency

This manuscript introduces a methodology (within the Born-Oppenheimer picture) to compute electronic ground-state properties of molecules and solids/surfaces with fractionally occupied components. Given a user-defined division of the molecule into subsystems, our theory uses an auxiliary global Hamiltonian that is defined as the sum of subsystem Hamiltonians, plus the spatial integral of a second-quantized local operator that allows the electrons to be transferred between subsystems. This electron transfer operator depends on a local potential that can be determined using density functional approximations and/or other techniques such as machine learning. The present framework employs superpositions of tensor-product wave functions, which can satisfy size consistency and avoid spurious fractional charges at large bond distances. The electronic population of each subsystem is in general a positive real number and is obtained from wave-function amplitudes, which are calculated by means of ground-state matrix diagonalization (or matrix propagation in the time-dependent case). Furthermore, our method can provide pathways to explore charge-transfer effects in environments where dividing the molecule into subsystems is convenient and to develop computationally affordable electronic structure algorithms.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Magnetic coupling in the disks around young gas giant planets

We examine the conditions under which the disks of gas and dust orbiting young gas giant planets are sufficiently conducting to experience turbulence driven by the magneto-rotational instability. By modeling the ionization and conductivity in the disk around proto-Jupiter, we find that turbulence is possible if the X-rays emitted near the Sun reach the planet's vicinity and either (1) the gas surface densities are in the range of the minimum-mass models constructed by augmenting Jupiter's satellites to solar composition, while dust is depleted from the disk atmosphere, or (2) the surface densities are much less, and in the range of gas-starved models fed with material from the solar nebula, but not so low that ambipolar diffusion decouples the neutral gas from the plasma. The results lend support to both minimum-mass and gas-starved models of the protojovian disk. (1) The dusty minimum-mass models have internal conductivities low enough to prevent angular momentum transfer by magnetic forces, as required for the material to remain in place while the satellites form. (2) The gas-starved models have magnetically active surface layers and a decoupled interior 'dead zone'. Similar active layers in the solar nebula yield accretion stresses in the range assumed in constructing the circumjovian gas-starved models. Our results also point to aspects of both classes of models that can be further developed. Non-turbulent minimum-mass models will lose dust from their atmospheres by settling, enabling gas to accrete through a thin surface layer. For the gas-starved models it is crucial to learn whether enough stellar X-ray and ultraviolet photons reach the circumjovian disk. Additionally, the stress-to-pressure ratio ought to increase with distance from the planet, likely leading to episodic accretion outbursts.

79 ASTRONOMY AND ASTROPHYSICS↗

MarCO: Interplanetary Mission Development on a CubeSat Scale

Shortly after JPL’s Interior Exploration using Seismic Investigations, Geodesy and Heat Transport (InSight) mission launches, separates, and commences its cruise phase, two CubeSats will deploy from the launch vehicle’s upper stage and begin independent flight to Mars (Fig. 1). During InSight’s entry, descent, and landing (EDL) sequence, these twin Mars Cube One (MarCO) spacecraft will fly 3,500 km above the Martian surface, recording and relaying InSight UHF radio data to the Deep Space Network (DSN) on Earth1. MarCO is a twin CubeSat mission developed by the NASA Jet Propulsion Laboratory (JPL) to accompany the InSight (Interior Exploration using Seismic Investigations, Geodesy and Heat Transport) Mars mission lander. MarCO's primary mission objective is to launch with InSight and independently fly to Mars to serve as a communications relay during InSight's entry, descent, and landing (EDL) phase. MarCO represents a new type of deep space mission: CubeSats at Mars. Building on the development of JPL's first interplanetary CubeSat project, the Interplanetary Nano-Spacecraft Pathfinder in Relevant Environment (INSPIRE), MarCO further refined the approach to hardware, software, and ground architecture development to solve the challenges of quickly building low-budget spacecraft to fly to Mars. The greatest constraint, beyond others typical of CubeSat missions, was time. The duration between MarCO's conception to completion of spacecraft assembly was less than two years - an unprecedented schedule for any planetary mission to date. Through necessity, MarCO has built on previous experience, procedures, systems, and development methodologies, defining a new niche for supporting larger primary missions. The MarCO spacecraft are poised to write a new chapter in deep space exploration. Originally slated to launch and reach Mars in 2016, the InSight mission schedule subsequently slipped to 2018. During the original landing of InSight, Earth would not be in view, and no orbiter around Mars would have been in position to both receive UHF EDL data and simultaneously relay it back to Earth. It was from this obstacle that MarCO was conceived. Regardless of any changes to InSight’s 2018 EDL configuration geometry, MarCO is still expected to fly and serve in the same capacity as originally designed: the first CubeSat mission to Mars. CubeSats have historically been firmly in the domain of universities and small companies. As first conceived, they served as a platform upon which to teach all aspects of the space mission lifecycle. JPL took on this mission type with Interplanetary Nano-Spacecraft Pathfinder in Relevant Environment2 (INSPIRE), moving the concept into a new domain: deep space. Building from the INSPIRE platform and lessons learned, MarCO addressed new challenges in the domain of planetary missions: independent interplanetary flight and navigation, integration with a large-scale mission, long-distance and long-delay communication, short development time, and a small development team. Of these, the greatest constraint was schedule: only 18 months passed from conception of mission concept until delivery of fully assembled and tested flight hardware. Careful selection of mission team, along with extensive use of off-the-shelf equipment, and streamlining automated processes, was essential. This achievement represents the next step in the evolution of CubeSats beyond low-Earth orbit.

Werne, Thomas↗

Interactive Machine Learning at Scale with CHISSL

We demonstrate CHISSL, a scalable client-server system for real-time interactive machine learning. Our system is capa- ble of incorporating user feedback incrementally and imme- diately without a structured or pre-defined prediction task. Computation is partitioned between a lightweight web-client and a heavyweight server. The server relies on representation learning and agglomerative clustering to learn a dendrogram, a hierarchical approximation of a representation space. The client uses only this dendrogram to incorporate user feedback into the model via transduction. Distances and predictions for each unlabeled instance are updated incrementally and deter- ministically, with O(n) space and time complexity. Our al- gorithm is implemented in a functional prototype, designed to be easy to use by non-experts. The prototype organizes the large amounts of data into recommendations. This allows the user to interact with actual instances by dragging and drop- ping to provide feedback in an intuitive manner. We applied CHISSL to several domains including cyber, social media, and geo-temporal analysis.

Arendt, Dustin L.↗

BraggNN : fast X-ray Bragg peak analysis using deep learning

X-ray diffraction based microscopy techniques such as high-energy diffraction microscopy (HEDM) rely on knowledge of the position of diffraction peaks with high precision. These positions are typically computed by fitting the observed intensities in detector data to a theoretical peak shape such as pseudo-Voigt. As experiments become more complex and detector technologies evolve, the computational cost of such peak-shape fitting becomes the biggest hurdle to the rapid analysis required for real-time feedback in experiments. To this end, we propose BraggNN, a deep-learning based method that can determine peak positions much more rapidly than conventional pseudo-Voigt peak fitting. When applied to a test dataset, peak center-of-mass positions obtained from BraggNN deviate less than 0.29 and 0.57 pixels for 75 and 95% of the peaks, respectively, from positions obtained using conventional pseudo-Voigt fitting (Euclidean distance). When applied to a real experimental dataset and using grain positions from near-field HEDM reconstruction as ground-truth, grain positions using BraggNN result in 15% smaller errors compared with those calculated using pseudo-Voigt. Recent advances in deep-learning method implementations and special-purpose model inference accelerators allow BraggNN to deliver enormous performance improvements relative to the conventional method, running, for example, more than 200 times faster on a consumer-class GPU card with out-of-the-box software.

36 MATERIALS SCIENCE↗

Multiclass Continuous Correspondence Learning

We extend the Structural Correspondence Learning (SCL) domain adaptation algorithm of Blitzer er al. to the realm of continuous signals. Given a set of labeled examples belonging to a 'source' domain, we select a set of unlabeled examples in a related 'target' domain that play similar roles in both domains. Using these 'pivot samples, we map both domains into a common feature space, allowing us to adapt a classifier trained on source examples to classify target examples. We show that when between-class distances are relatively preserved across domains, we can automatically select target pivots to bring the domains into correspondence.

correspondence learning↗

Genarris 2.0: A Random Structure Generator for Molecular Crystals

Genarris is an open source Python package for generating random molecular crystal structures with physical constraints for seeding crystal structure prediction algorithms and training machine learning models. Here we present a new version of the code, containing several major improvements. A MPI-based parallelization scheme has been implemented, which facilitates the seamless sequential execution of user-defined workflows. A new method for estimating the unit cell volume based on the single molecule structure has been developed using a machine-learned model trained on experimental structures. A new algorithm has been implemented for generating crystal structures with molecules occupying special Wyckoff positions. A new hierarchical structure check procedure has been developed to detect unphysical close contacts efficiently and accurately. New intermolecular distance settings have been implemented for strong hydrogen bonds. To demonstrate these new features, we study two specific cases: benzene and glycine. Genarris finds the experimental structures of the two polymorphs of benzene and the three polymorphs of glycine. Program summary Program Title: Genarris 2.0 Program Files doi: http://dx.doi.org/10.17632/grx6mz4pjn.1 Licensing provisions: BSD-3 Clause Programming language: Python, C External routines/libraries: Spglib, ASE, pymatgen, SciPy, mpi4py, scikit-learn, PyTorch, FHI-aims. Nature of problem: Molecular crystal structure prediction. Solution method: Genarris 2.0 generates molecular crystal structures over the 230 space groups, on general and special Wyckoff positions, using physical constraints. Down-sampling of the generated structures may be performed subsequently, based on molecular crystal packing descriptors and an unsupervised machine learning algorithm. Lastly, ab initio structure relaxation may be performed for the final pool. Depending on the user-defined workflow implemented, Genarris may be used to generate diverse molecular crystal datasets to seed evolutionary algorithms or to train machine learning algorithms or as a standalone crystal structure prediction method. Restrictions: For crystal structure generation, the molecule of interest must be semi-rigid with no bond rotational degrees of freedom. Unusual features: Genarris 2.0 is a highly distributed program, making use of MPI for Python parallelization. The user has the ability to design and implement workflows by executing a user-defined list of procedures. Genarris 2.0 offers new features including a machine learning model for estimating the molecular volume in the solid state from the single molecule structure, structure generation in special Wyckoff positions of space groups, hierarchical structure checks including rigorous treatment of non-orthogonal structures, and clustering and down-selection workflows combining first principles simulations with machine learning. (C) 2020 Elsevier B.V. All rights reserved.

Crystal structure prediction↗

Chasing Accreted Structures within Gaia DR2 Using Deep Learning

In previous work, we developed a deep neural network classifier that only relies on phase-space information to obtain a catalog of accreted stars based on the second data release of Gaia (DR2). In this paper, we apply two clustering algorithms to identify velocity substructure within this catalog. We focus on the subset of stars with line-of-sight velocity measurements that fall in the range of Galactocentric radii $r\in [6.5,9.5]\,{\rm{kpc}}$ and vertical distances $| z| \lt 3\,{\rm{kpc}}$. Known structures such as Gaia Enceladus and the Helmi stream are identified. The largest previously unknown structure, Nyx, is a vast stream consisting of at least 200 stars in the region of interest. This study displays the power of the machine-learning approach by not only successfully identifying known features but also discovering new kinematic structures that may shed light on the merger history of the Milky Way.

Astronomy & Astrophysics↗

Recent Advances in Machine Learning for Fiber Optic Sensor Applications

Over the last three decades, fiber optic sensors (FOS) have gained a lot of attention for their wide range of monitoring applications across many industries, including aerospace, defense, security, civil engineering, and energy. FOS technologies hold great promise to form the backbone for next‐generation intelligent sensing platforms that offer long‐distance, high‐accuracy, distributed measurement capabilities and multiparametric monitoring with resilience to harsh environmental conditions. The major limitations posed by FOS are 1) cross‐sensitivity, 2) enormous volume and large data generation, 3) low data processing speed, 4) degradation of signal‐to‐noise ratio over the fiber length, and 5) overall cost of sensor and interrogator systems. These challenges can be overcome by building advanced data analytics engines enabled by recent breakthroughs in machine learning (ML) and artificial intelligence (AI). This article presents a comprehensive review of recent studies that integrate ML and AI algorithms with FOS technologies. This review also highlights several FOS technology development directions that promise a significant impact on widespread use for several industrial applications, with an emphasis on energy systems monitoring. A perspective on future directions for further research development is also provided.

97 MATHEMATICS AND COMPUTING↗

DiffLense: a conditional diffusion model for super-resolution of gravitational lensing data

Abstract Gravitational lensing data is frequently collected at low resolution due to instrumental limitations and observing conditions. Machine learning-based super-resolution techniques offer a method to enhance the resolution of these images, enabling more precise measurements of lensing effects and a better understanding of the matter distribution in the lensing system. This enhancement can significantly improve our knowledge of the distribution of mass within the lensing galaxy and its environment, as well as the properties of the background source being lensed. Traditional super-resolution techniques typically learn a mapping function from lower-resolution to higher-resolution samples. However, these methods are often constrained by their dependence on optimizing a fixed distance function, which can result in the loss of intricate details crucial for astrophysical analysis. In this work, we introduce DiffLense , a novel super-resolution pipeline based on a conditional diffusion model specifically designed to enhance the resolution of gravitational lensing images obtained from the Hyper Suprime-Cam Subaru Strategic Program (HSC-SSP). Our approach adopts a generative model, leveraging the detailed structural information present in Hubble space telescope (HST) counterparts. The diffusion model, trained to generate HST data, is conditioned on HSC data pre-processed with denoising techniques and thresholding to significantly reduce noise and background interference. This process leads to a more distinct and less overlapping conditional distribution during the model’s training phase. We demonstrate that DiffLense outperforms existing state-of-the-art single-image super-resolution techniques, particularly in retaining the fine details necessary for astrophysical analyses.

Computer Science↗

Data Locality Enhancement of Dynamic Simulations for Exascale Computing (Final Report)

The development of modern processors exhibits two trends that complicate the optimizations of modern software. The first is the increasing sensitivity of processors' throughput to irregularities in computation. With more processors produced through a massive integration of simple cores, future systems will increasingly favor regular data-level parallel computations, but deviate from the needs of applications with complex patterns. Some evidences are already shown on Graphic Processing Units (GPU): Irregular data accesses (e.g., indirect references A[D[i]]) and conditional branches are limiting many GPU applications' performance at a level an order of magnitude lower than the peak of GPU. The second hardware trend is the growing gap between memory bandwidth and the aggregate speed—that is, the sum of all cores' computing power—of a Chip Multiprocessor (CMP). Despite the capped growth of the peak CPU speed, the aggregate speed of a CMP keeps increasing as more cores get into a single chip. It is expected that by 2018, node concurrency in an exascale system will increase by hundreds of times, whereas, memory bandwidth will expand by only 10 to 20 times. Consequently, data movement and storage is expected to consume more than 70% of the total system power. Bridging this gap is difficult; the complexities of modern CMP memory hierarchy make it even harder: Data cache becomes shared among computing units, and the sharing is often non-uniform—whether two computing units share a cache depends on their proximity and the level of the cache. On the recent IBM Power7 architecture, for instance, four hardware contexts (or SMT threads) in a core share the entire memory hierarchy, all cores in one chip share an on-chip L3 cache, and cores across chips share L3 and main memory through off-chip connections. These two trends complicate the translation of computing power into performance, especially for a program with either intensive data accesses or complex patterns in data accesses or control flow paths. Unfortunately, both attributes present and will persist in a class of important applications. For instance, many scientific simulations deal with a large volume of data. And meanwhile, as most real-world processes are non-uniform and evolving (e.g., the evolution of a galaxy or the process of a drug injection), both the computations and data accesses of these programs tend to be irregular and dynamically changing. Currently, the lack of support to these applications on modern CMP severely limits their performance. On GPU, as our recent study shows and other studies echo, performance enhancement of a factor of integers is possible when memory accesses or control flows are streamlined for a set of GPU applications. On multicore CPU, our studies show that traditional locality enhancement, for being oblivious to the new features of multicore memory hierarchy, may even cause large slowdown to data-intensive dynamic applications. The severity of the issues is expected to worsen as the two hardware trends continue. Some recent studies try to match software with the trends, but in a limited scope or manner. For irregularities on GPU, most studies focus on irregularities analyzable through static analysis (e.g., data accesses in regular loops). Dynamic irregularities are harder to address because the needed analysis and transformations typically have to happen at run time. Some other research resorts to hardware extensions, an actual adoption of which is unclear for the entailed space cost and complexity. For data locality, recent years have seen some exploitations of the new memory hierarchy on multicore for performance, but most of them are on process or thread scheduling, rather than program transformations. Our recent study reveals that program-level transformations may magnify the scheduling benefits by a factor of seven, concluding that program-level transformation should play a central role for data locality enhancement on modern CMP. But research in this direction has been sparse, and most have focused on data layout or cache performance modeling, rather than program transformations to match with the new memory hierarchy features. Overall, it is still an open question how to bridge the gap between dynamic computations and the two prominent properties of modern processors. The goal of this project is to develop a set of techniques and software tools to enhance the matching between memory accesses in dynamic simulations and the prominent features of modern and future CMP, alleviating the memory performance issues for petascale and exascale computing. This report summarizes the discoveries and products produced throughout this project. It includes free launch, a new software approach to overcoming the shortcomings of both methods; coherence-free multiview, an approach that allows multiple views of a single data object to co-exist on GPU memory during a GPU kernel execution; algorithmic optimizations to data analytics problems, especially those that involve lots of distance calculations. Supporting dynamic parallelism is important for GPU to benefit a broad range of applications. There are currently two fundamental ways for programs to exploit dynamic parallelism on GPU: a software-based approach with software-managed worklists, and a hardware-based approach through dynamic subkernel launches. Neither is satisfactory. The former is complicated to program and is often subject to some load imbalance; the latter suffers large runtime overhead. In this work, we propose free launch, a new software approach to overcoming the shortcomings of both methods. It allows programmers to use subkernel launches to express dynamic parallelism. It employs a novel compiler-based code transformation named subkernel launch removal to replace the subkernel launches with the reuse of parent threads. Coupled with an adaptive task assignment mechanism, the transformation reassigns the tasks in the subkernels to the parent threads with a good load balance. The technique requires no hardware extensions, immediately deployable on existing GPUs. It keeps the programming convenience of the subkernel launch-based approach while avoiding its large runtime overhead. Meanwhile, its superior load balancing makes it outperform manual worklist-based techniques by 3X on average. The work was published at the 48th Annual IEEE/ACM International Symposium on Microarchitecture (Micro'2015). A Graphic Processing Unit (GPU) system is typically equipped with many types of memory (e.g., global, constant, texture, shared, cache). Data placement determines what data are placed on which type of memory, essential for GPU memory performance. Prior optimizations of data placement always require a single view of a data object on memory, which limits the optimization effectiveness. In this work, we propose coherence-free multiview, an approach that allows multiple views of a single data object to co-exist on GPU memory during a GPU kernel execution. We demonstrate that under certain conditions, the multiple views can remain incoherent while facilitating enhanced data placement. We present a theorem and some compiler support to ensure the soundness of the usage of coherence-free multiview. We further develop reference-discerning data placement, a new way to enhance data placements on GPU. It enables more flexible data placements by using coherence-free multiview to leverage the slack in coherence requirement of some GPU programs. Experiments on three types of GPU systems show that, with less than 200KB space cost, the new data placement technique can provide a 1.6X average (up to 4.27X) speedup. In addition, we have examined algorithmic optimizations to data analytics problems, especially those that involve lots of distance calculations. Computing distances among data points is an essential part of many important algorithms in data analytics, graph analysis, and other domains. In each of these domains, developers have spent significant manual effort optimizing algorithms, often through novel applications of the triangle equality, in order to minimize the number of distance computations in the algorithms. In this work, we observe that many algorithms across these domains can be generalized as an instance of a generic distance-related abstraction. Based on this abstraction, we derive seven principles for correctly applying the triangular inequality to optimize distance-related algorithms. Guided by the findings, we develop Triangular OPtimizer (TOP), the first software framework that is able to automatically produce optimized algorithms that either matches or outperforms manually designed algorithms for solving distance-related problems. TOP achieves up to 237x speedups and 2.5X on average. The work has been published at the 32nd International Conference on Machine Learning (ICML'15) and the 41st International Conference on Very Large Data Bases (VLDB'15).

42 ENGINEERING↗

Process Optimization of Carbon Electrode Materials Manufacturing by Experimental Study and Machine Learning Techniques

Electrospun carbon fibers from coal have been investigated as electrodes for batteries and supercapacitors. Despite the excellent properties of coal-derived carbon fibers (CCNF) for energy storage devices, there still lacks systematic understanding on how various process parameters affect final electrode performances, which poses challenges to scale from pilot to high volume manufacturing. The goals of this project are twofold. First, we focuse on process optimization for converting a new precursor from powder river basin (PRB) coal, referred to as coal-based polyurethane (CPU) to CCNF using electrospinning. Second, different machine learning techniques will be examined using experimental data from this work and open literature. Specifically, for CPU the following process parameters need to be characterized and optimized in order to produce CCNFs with desirable mechanical integrity and physiochemical properties: precursor composition and viscosity, operating voltage and distance, oxidation and carbonization temperature and duration. Consequently, physiochemical properties of the fibers were characterized to correlate these process parameters with desirable electrochemical performance. Given the complex nature of the fiber production process, ML models are assessed for their ability to capture the nonlinear relationship between process parameters and the electrochemical properties in applications including supercapacitors. As such, we applied various machine learning techniques, to determine which technique produces a model that best predicts device function.

Cincotta, Robert E.F.↗