SEARCH · Search NASA

Results for “supervised training”

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.

SourceResearch connectionFrom yearThrough year

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

At least 19 records

All sources NASA NTRS DOE OSTI

NASA NTRSDate not supplied

Towards an Aviation Large Language Model by Fine-tuning and Evaluating Transformers

In the aviation domain, there are many applications for machine learning and artificial intelligence tools that utilize natural language. For example, there is a desire to know the commonalities in written safety reports such as voluntary post incidents reports or aerial wildfire operations reports to better understand the risks present. Another use-case is the possibility of extracting airspace procedures and constraints currently written in documents such as Letters of Agreement. These applications can benefit from the use of state-of-the-art natural language processing techniques when adapted to the language/phraseology specific to the aviation domain. This paper evaluates the viability of adaptation of NLP tools to the aviation domain by fine-tuning transformer based models using aviation data sets. In 2018, a novel language model based on neural units (also called transformers) was created and became known as “Bidirectional Encoder Representations from Transformers” or BERT. This architecture combined with large amounts of English training data and innovative semi-supervised training tasks set the standard for what would later emerge as Large Language Models. The performance of these models was further improved by hyperparameter tuning and refinement of the semi-supervised training task and resulted in “Robustly Optimized BERT Pre-training Approach through hyperparameter tuning” or RoBERTa models. These pre-trained Large Language Models proved to be useful for a wide variety of natural language processing tasks such as text classification and question answering through a process called fine-tuning. The transformer architecture with pre-trained weights served as the basis with the last few layers replaced with layers fine-tuned to perform a new task e.g., a layer that provides a label for the entire input text. This process of fine-tuning can also be used to adapt the models to new domains; e.g., BioBERT started with the pre-trained BERT model and was completed by additional fine-tuning and training on biomedical documents. Transformer-based architectures can also be used to create rich representations of text called embeddings which can serve as the input to other machine learning models. This allows simpler algorithms such as logistic regression to use context-rich representations of the text while still remaining quick to train and evaluate. In the world of aviation, there is a growing demand for natural language processing and understanding but the domain presents unique challenges. Due to the technical content (and specialized language) of most aviation documents, fine-tuning pre-trained Large Language Models to specific tasks has not met the benchmark on natural language processing tasks set by simpler models trained from scratch on the data. To address this deficiency, this paper evaluates the improvements from fine-tuning a Large Language Model on a large set of aviation documents using the original semi-supervised training tasks before performing specific natural language tasks. In fine-tuning, a domain-specific dataset is used on the original training task but with the pre-trained Large Language Model instead of starting from a random initialization. This approach allows the model to be adapted to the specific domain language without discarding the information gained from training on general English data. This paper utilized two major dataset types to train and assess the RoBERTa fine-tuning performance. The first are 7,057 Letters of Agreement which are Federal Aviation Administration (FAA) documents that formalize airspace operations across the national airspace system. They contain many examples of ‘aviation English’ using domain specific terminology and phrasing which serves as a representative basis to perform the semi-supervised fine-tuning. The second type is the 494 document classification labels to be used for evaluation. This down-stream evaluation aims to show the performance of the fine-tuned model, better understand how much data is needed for an effective fine-tuning, and how fine-tuning can be adapted for different applications in-the domain. After semi-supervised training, evaluation begins by encoding the documents for classification using the fine-tuned RoBERTa model. Then a logistic regression classifier is trained to label the document type and compared against our ground truth labels. This currently leads to a 82.8% accuracy on 10-fold cross validation showing improvement over baseline RoBERTa which achieved 81.0%. We plan to measure the improvements on additional tasks and it is expected that these improvements will lead to more robust models that can tackle the natural language processing challenges present in aviation datasets.

ATM↗

NASA NTRS1994

Classification of asteroid spectra using a neural network

The 52-color asteroid survey (Bell et al., 1988) together with the 8-color asteroid survey (Zellner et al., 1985) provide a data set of asteroid spectra spanning 0.3-2.5 micrometers. An artificial neural network clusters these asteroid spectra based on their similarity to each other. We have also trained the neural network with a categorization learning output layer in a supervised mode to associate the established clusters with taxonomic classes. Results of our classification agree with Tholen's classification based on the 8-color data alone. When extending the spectral range using the 52-color survey data, we find that some modification of the Tholen classes is indicated to produce a cleaner, self-consistent set of taxonomic classes. After supervised training using our modified classes, the network correctly classifies both the training examples, and additional spectra into the correct class with an average of 90% accuracy. Our classification supports the separation of the K class from the S class, as suggested by Bell et al. (1987), based on the near-infrared spectrum. We define two end-member subclasses which seem to have compositional significance within the S class: the So class, which is olivine-rich and red, and the Sp class, which is pyroxene-rich and less red. The remaining S-class asteroids have intermediate compositions of both olivine and pyroxene and moderately red continua. The network clustering suggests some additional structure within the E-, M-, and P-class asteroids, even in the absence of albedo information, which is the only discriminant between these in the Tholen classification. New relationships are seen between the C class and related G, B, and F classes. However, in both cases, the number of spectra is too small to interpret or determine the significance of these separations.

Howell, E. S.↗

NASA NTRS1983

LANDSAT-4 image data quality analysis

Seven heterogeneous areas within the entire Des Moines, Iowa test site were selected to define candidate spectral training classes using a clustering algorithm. In addition to the 91 cluster nonsupervised classes, three supervised training classes were defined. The original candidate training classes were reduced to 42 spectrally separable training classes. The minimum and average transformed divergence values for the 42 spectral classes and for the best subsets of Y TM spectral bands are shown in a table. The best spectral band for any combination of 1 through 7 bands is the first middle IR band. The next best band is the near IR, followed by the red band and than the thermal IR. The best combination of four bands includes one from each of the four regions of the spectrum (visible, near IR, middle IR, and thermal IR).

Anuta, P. E.↗

NASA NTRS1983

LANDSAT-4 image data quality analysis

Seven heterogeneous areas within the Des Moines, Iowa area test site were selected to define candidate spectral training classes using a clustering algorithm. In addition to the 91 cluster (nonsupervised) classes, three supervised training classes were defined and subsequently included in the training statistics file. The identity of all 94 candidate classes were determined using available reference data. Through analysis of the interclass separabilities, the original 94 candidate training classes were reduced to 42 spectrally separable final classes. The minimum average transformed divergence values for the 42 spectral classes and for the best subsets of TM spectral bands are shown in a table.

Anuta, P. E.↗

NASA NTRSDate not supplied

Developing Natural Language Processing and Supervised Learning Techniques to Classify Mars Tasks

As NASA's Human Research Program (HRP) prepares for long-duration Mars missions, understanding astronaut tasks is crucial. This study, conducted at NASA Glenn Research Center (GRC), employed Natural Language Processing (NLP) and machine learning techniques to analyze and classify Mars tasks. A list of 1,058 Mars tasks was provided by HRP experts including binary labeling of 18 Human System Task Categories (HSTCs). We developed an NLP model using Google's BERT language model to capture the semantic and syntactic nuances of these tasks. Supervised training was initially applied to a subset of the NLP-analyzed tasks to assess the model's effectiveness in classifying the remaining tasks. Incorporating HSTC descriptions significantly enhanced the classification accuracy for 9 out of the 18 HSTCs and reduced training time. To address the issue of severe class imbalance in the HSTC data, we introduced innovative weighting and sampling techniques for data augmentation. We then fine-tune BERT to implement a pairwise relatedness scoring method, allowing us to cluster tasks based on their relatedness and similarity, getting a step closer to labeling the tasks without supervision. In this presentation we guide you through data preprocessing, deciphering key syntax components using BERT, and performing supervised classification of the Mars tasks. This work showcases the potential use of advanced NLP techniques to analyze Mars missions to be incorporated into various crew health and performance analyses.

GenAI↗

NASA NTRSDate not supplied

Improving Sim-to-Real Transfer in Vision-Based Robot Navigation Via Instance-Level GAN-Based Data Augmentation

Achieving robust vision-based robotic tasks requires large amounts of data, which are often difficult to obtain in real-world scenarios. Simulators and synthetic data offer a cost-effective alternative, but the visual gap between simulation and reality hinders the performance of models when deployed in real-world environments. In this paper, we present a data augmentation pipeline that integrates a foundation model (Segment Anything Model) with an unsupervised image-to-image translation model (CycleGAN) for instance-level domain transfer from simulation to reality. This pipeline enables the generation of realistic labeled data from synthetic images for training supervised machine learning models in vision-based navigation tasks. We evaluate our approach on real-world data for ego-vehicle pose estimation, a critical autonomous navigation task involving the prediction of cross-track position and heading angle relative to road center line markings. The results of our tests show that our GAN-based data augmentation pipeline significantly outperforms models trained solely on simulation data or on data processed with standard image augmentation methods for sim-to-real transfer, enhancing model robustness and generalizability in real-world scenarios. Our method provides a scalable and flexible data augmentation tool for leveraging large synthetic datasets to enhance vision-based robotic navigation tasks.

artificial intelligence↗

NASA NTRS1970

Preprocessing of multispectral data and simulation of ERTS data channels to make computer terrain maps of a Yellowstone National Park test site

The possibility of improving the accuracy of terrain classification by preprocessing spectral data was investigated. Terrain maps were made using the following techniques: 1) preprocessing by scan angle function transformation, using the computer-selected best set of three channels; and 2) preprocessing by ratio transformation, using the specified ERTS data channels, simulated by fitting the spectral response of each of the 12 data channels to the ERTS channels by a set of weighting coefficients. By using a simple technique during printout, the maps were produced in color. The normalized scan angle function transformation resulted in the most accurate classification. The best ratio transformation for the Yellowstone Park data was the ratio of each channel to the sum of all channels. A supervised training program involving maximum likelihood decision for selecting the best spectrometer channels and similar techniques for digitizing the data of the analog magnetic tapes were used. Cloud shadows were recognized in addition to eight classes of terrain. Preprocessing of data resulted in more accurate maps, required fewer training areas (hence less preparation and computer time), and enabled much of the area formerly classified as shadow to be reclassified according to actual terrain type.

Smedes, H. W.↗

NASA NTRS1980

Development of alternative data analysis techniques for improving the accuracy and specificity of natural resource inventories made with digital remote sensing data

An investigation was conducted into ways to improve the involvement of state and local user personnel in the digital image analysis process by isolating those elements of the analysis process which require extensive involvement by field personnel and providing means for performing those activities apart from a computer facility. In this way, the analysis procedure can be converted from a centralized activity focused on a computer facility to a distributed activity in which users can interact with the data at the field office level or in the field itself. A general image processing software was developed on the University of Minnesota computer system (Control Data Cyber models 172 and 74). The use of color hardcopy image data as a primary medium in supervised training procedures was investigated and digital display equipment and a coordinate digitizer were procured.

Lillesand, T. M.↗

NASA NTRS1984

An automated land-use mapping comparison of the Bayesian maximum likelihood and linear discriminant analysis algorithms

The Bayesian maximum likelihood parametric classifier has been tested against the data-based formulation designated 'linear discrimination analysis', using the 'GLIKE' decision and "CLASSIFY' classification algorithms in the Landsat Mapping System. Identical supervised training sets, USGS land use/land cover classes, and various combinations of Landsat image and ancilliary geodata variables, were used to compare the algorithms' thematic mapping accuracy on a single-date summer subscene, with a cellularized USGS land use map of the same time frame furnishing the ground truth reference. CLASSIFY, which accepts a priori class probabilities, is found to be more accurate than GLIKE, which assumes equal class occurrences, for all three mapping variable sets and both levels of detail. These results may be generalized to direct accuracy, time, cost, and flexibility advantages of linear discriminant analysis over Bayesian methods.

Tom, C. H.↗

NASA NTRS1991

Parameter estimation in space systems using recurrent neural networks

The identification of time-varying parameters encountered in space systems is addressed, using artificial neural systems. A hybrid feedforward/feedback neural network, namely a recurrent multilayer perception, is used as the model structure in the nonlinear system identification. The feedforward portion of the network architecture provides its well-known interpolation property, while through recurrency and cross-talk, the local information feedback enables representation of temporal variations in the system nonlinearities. The standard back-propagation-learning algorithm is modified and it is used for both the off-line and on-line supervised training of the proposed hybrid network. The performance of recurrent multilayer perceptron networks in identifying parameters of nonlinear dynamic systems is investigated by estimating the mass properties of a representative large spacecraft. The changes in the spacecraft inertia are predicted using a trained neural network, during two configurations corresponding to the early and late stages of the spacecraft on-orbit assembly sequence. The proposed on-line mass properties estimation capability offers encouraging results, though, further research is warranted for training and testing the predictive capabilities of these networks beyond nominal spacecraft operations.

Parlos, Alexander G.↗

NASA NTRS1992

Neural network based decomposition in optimal structural synthesis

The present paper describes potential applications of neural networks in the multilevel decomposition based optimal design of structural systems. The generic structural optimization problem of interest, if handled as a single problem, results in a large dimensionality problem. Decomposition strategies allow for this problem to be represented by a set of smaller, decoupled problems, for which solutions may either be obtained with greater ease or may be obtained in parallel. Neural network models derived through supervised training, are used in two distinct modes in this work. The first uses neural networks to make available efficient analysis models for use in repetitive function evaluations as required by the optimization algorithm. In the second mode, neural networks are used to represent the coupling that exists between the decomposed subproblems. The approach is illustrated by application to the multilevel decomposition-based synthesis of representative truss and frame structures.

Hajela, P.↗

NASA NTRS1996

Improving the Performance of the Structure-Based Connectionist Network for Diagnosis of Helicopter Gearboxes

A diagnostic method is introduced for helicopter gearboxes that uses knowledge of the gear-box structure and characteristics of the 'features' of vibration to define the influences of faults on features. The 'structural influences' in this method are defined based on the root mean square value of vibration obtained from a simplified lumped-mass model of the gearbox. The structural influences are then converted to fuzzy variables, to account for the approximate nature of the lumped-mass model, and used as the weights of a connectionist network. Diagnosis in this Structure-Based Connectionist Network (SBCN) is performed by propagating the abnormal vibration features through the weights of SBCN to obtain fault possibility values for each component in the gearbox. Upon occurrence of misdiagnoses, the SBCN also has the ability to improve its diagnostic performance. For this, a supervised training method is presented which adapts the weights of SBCN to minimize the number of misdiagnoses. For experimental evaluation of the SBCN, vibration data from a OH-58A helicopter gearbox collected at NASA Lewis Research Center is used. Diagnostic results indicate that the SBCN is able to diagnose about 80% of the faults without training, and is able to improve its performance to nearly 100% after training.

Jammu, Vinay B.↗

NASA NTRS2017

Development of Increasingly Autonomous Traffic Data Manager Using Pilot Relevancy and Ranking Data

NASA's Safe Autonomous Systems Operations (SASO) project goal is to define and safely enable all future airspace operations by justifiable and optimal autonomy for advanced air, ground, and connected capabilities. This work showcases how Increasingly Autonomous Systems (IAS) could create operational transformations beneficial to the enhancement of civil aviation safety and efficiency. One such IAS under development is the Traffic Data Manager (TDM). This concept is a prototype 'intelligent party-line' system that would declutter and parse out non-relevant air traffic, displaying only relevant air traffic to the aircrew in a digital data communications (Data Comm) environment. As an initial step, over 22,000 data points were gathered from 31 Airline Transport Pilots to train the machine learning algorithms designed to mimic human experts and expertise. The test collection used an analog of the Navigation Display. Pilots were asked to rate the relevancy of the displayed traffic using an interactive tablet application. Pilots were also asked to rank the order of importance of the information given, to better weight the variables within the algorithm. They were also asked if the information given was enough data, and more importantly the "right" data to best inform the algorithm. The paper will describe the findings and their impact to the further development of the algorithm for TDM and, in general, address the issue of how can we train supervised machine learning algorithms, critical to increasingly autonomous systems, with the knowledge and expertise of expert human pilots.

Le Vie, Lisa R.↗

NASA NTRS2011

Overview of Pre-Flight Physical Training, In-Flight Exercise Countermeasures and the Post-Flight Reconditioning Program for International Space Station Astronauts

International Space Station (ISS) astronauts receive supervised physical training pre-flight, utilize exercise countermeasures in-flight, and participate in a structured reconditioning program post-flight. Despite recent advances in exercise hardware and prescribed exercise countermeasures, ISS crewmembers are still found to have variable levels of deconditioning post-flight. This presentation provides an overview of the astronaut medical certification requirements, pre-flight physical training, in-flight exercise countermeasures, and the post-flight reconditioning program. Astronauts must meet medical certification requirements on selection, annually, and prior to ISS missions. In addition, extensive physical fitness testing and standardized medical assessments are performed on long duration crewmembers pre-flight. Limited physical fitness assessments and medical examinations are performed in-flight to develop exercise countermeasure prescriptions, ensure that the crewmembers are physically capable of performing mission tasks, and monitor astronaut health. Upon mission completion, long duration astronauts must re-adapt to the 1 G environment, and be certified as fit to return to space flight training and active duty. A structured, supervised postflight reconditioning program has been developed to prevent injuries, facilitate re-adaptation to the 1 G environment, and subsequently return astronauts to training and space flight. The NASA reconditioning program is implemented by the Astronaut Strength, Conditioning, and Rehabilitation (ASCR) team and supervised by NASA flight surgeons. This program has evolved over the past 10 years of the International Space Station (ISS) program and has been successful in ensuring that long duration astronauts safely re-adapt to the 1 g environment and return to active duty. Lessons learned from this approach to managing deconditioning can be applied to terrestrial medicine and future exploration space flight missions.

Kerstman, Eric↗

NASA NTRS1976

DHARMA - Discriminant hyperplane abstracting residuals minimization algorithm for separating clusters with fuzzy boundaries

Learning of discriminant hyperplanes in imperfectly supervised or unsupervised training sample sets with unreliably labeled samples along the fuzzy joint boundaries between sample clusters is discussed, with the discriminant hyperplane designed to be a least-squares fit to the unreliably labeled data points. (Samples along the fuzzy boundary jump back and forth from one cluster to the other in recursive cluster stabilization and are considered unreliably labeled.) Minimization of the distances of these unreliably labeled samples from the hyperplanes does not sacrifice the ability to discriminate between classes represented by reliably labeled subsets of samples. An equivalent unconstrained linear inequality problem is formulated and algorithms for its solution are indicated. Landsat earth sensing data were used in confirming the validity and computational feasibility of the approach, which should be useful in deriving discriminant hyperplanes separating clusters with fuzzy boundaries, given supervised training sample sets with unreliably labeled boundary samples.

Dasarathy, B. V.↗

NASA NTRS1990

Artificial neural network classification using a minimal training set - Comparison to conventional supervised classification

Recent research has shown an artificial neural network (ANN) to be capable of pattern recognition and the classification of image data. This paper examines the potential for the application of neural network computing to satellite image processing. A second objective is to provide a preliminary comparison and ANN classification. An artificial neural network can be trained to do land-cover classification of satellite imagery using selected sites representative of each class in a manner similar to conventional supervised classification. One of the major problems associated with recognition and classifications of pattern from remotely sensed data is the time and cost of developing a set of training sites. This reseach compares the use of an ANN back propagation classification procedure with a conventional supervised maximum likelihood classification procedure using a minimal training set. When using a minimal training set, the neural network is able to provide a land-cover classification superior to the classification derived from the conventional classification procedure. This research is the foundation for developing application parameters for further prototyping of software and hardware implementations for artificial neural networks in satellite image and geographic information processing.

Hepner, George F.↗

NASA NTRS1994

Neural Network Classifies Teleoperation Data

Prototype artificial neural network, implemented in software, identifies phases of telemanipulator tasks in real time by analyzing feedback signals from force sensors on manipulator hand. Prototype is early, subsystem-level product of continuing effort to develop automated system that assists in training and supervising human control operator: provides symbolic feedback (e.g., warnings of impending collisions or evaluations of performance) to operator in real time during successive executions of same task. Also simplifies transition between teleoperation and autonomous modes of telerobotic system.

Fiorini, Paolo↗

NASA NTRS1998

Experimental Evaluation of a Structure-Based Connectionist Network for Fault Diagnosis of Helicopter Gearboxes

This paper presents the experimental evaluation of the Structure-Based Connectionist Network (SBCN) fault diagnostic system introduced in the preceding article. For this vibration data from two different helicopter gearboxes: OH-58A and S-61, are used. A salient feature of SBCN is its reliance on the knowledge of the gearbox structure and the type of features obtained from processed vibration signals as a substitute to training. To formulate this knowledge, approximate vibration transfer models are developed for the two gearboxes and utilized to derive the connection weights representing the influence of component faults on vibration features. The validity of the structural influences is evaluated by comparing them with those obtained from experimental RMS values. These influences are also evaluated ba comparing them with the weights of a connectionist network trained though supervised learning. The results indicate general agreement between the modeled and experimentally obtained influences. The vibration data from the two gearboxes are also used to evaluate the performance of SBCN in fault diagnosis. The diagnostic results indicate that the SBCN is effective in directing the presence of faults and isolating them within gearbox subsystems based on structural influences, but its performance is not as good in isolating faulty components, mainly due to lack of appropriate vibration features.

Jammu, V. B.↗