Search NASA⌕ Search

SEARCH · Search NASA

Results for “Decoder Input”

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 37 records · Page 2

Soft-Decision-Data Reshuffle to Mitigate Pulsed Radio Frequency Interference Impact on Low-Density-Parity-Check Code Performance

This presentation briefly discusses a research effort on mitigation techniques of pulsed radio frequency interference (RFI) on a Low-Density-Parity-Check (LDPC) code. This problem is of considerable interest in the context of providing reliable communications to the space vehicle which might suffer severe degradation due to pulsed RFI sources such as large radars. The LDPC code is one of modern forward-error-correction (FEC) codes which have the decoding performance to approach the Shannon Limit. The LDPC code studied here is the AR4JA (2048, 1024) code recommended by the Consultative Committee for Space Data Systems (CCSDS) and it has been chosen for some spacecraft design. Even though this code is designed as a powerful FEC code in the additive white Gaussian noise channel, simulation data and test results show that the performance of this LDPC decoder is severely degraded when exposed to the pulsed RFI specified in the spacecraft s transponder specifications. An analysis work (through modeling and simulation) has been conducted to evaluate the impact of the pulsed RFI and a few implemental techniques have been investigated to mitigate the pulsed RFI impact by reshuffling the soft-decision-data available at the input of the LDPC decoder. The simulation results show that the LDPC decoding performance of codeword error rate (CWER) under pulsed RFI can be improved up to four orders of magnitude through a simple soft-decision-data reshuffle scheme. This study reveals that an error floor of LDPC decoding performance appears around CWER=1E-4 when the proposed technique is applied to mitigate the pulsed RFI impact. The mechanism causing this error floor remains unknown, further investigation is necessary.

Ni, Jianjun David↗

A large-scale interactive one-dimensional array processing system

The work describes a scientist/user oriented interactive program for processing one-dimensional arrays. It is shown that the program is oriented toward processing spectrophotometric astronomical data and can also be used for general I-D array processing. Further, the program has totally free format input with a sophisticated decoding capability which can cope with typographical plus other possible mistakes. Finally, a description of the program is given to provide information on implementing a large-scale data-reduction facility.

Clark, R. N.↗

Error coding simulations in C

When data is transmitted through a noisy channel, errors are produced within the data rendering it indecipherable. Through the use of error control coding techniques, the bit error rate can be reduced to any desired level without sacrificing the transmission data rate. The Astrionics Laboratory at Marshall Space Flight Center has decided to use a modular, end-to-end telemetry data simulator to simulate the transmission of data from flight to ground and various methods of error control. The simulator includes modules for random data generation, data compression, Consultative Committee for Space Data Systems (CCSDS) transfer frame formation, error correction/detection, error generation and error statistics. The simulator utilizes a concatenated coding scheme which includes CCSDS standard (255,223) Reed-Solomon (RS) code over GF(2(exp 8)) with interleave depth of 5 as the outermost code, (7, 1/2) convolutional code as an inner code and CCSDS recommended (n, n-16) cyclic redundancy check (CRC) code as the innermost code, where n is the number of information bits plus 16 parity bits. The received signal-to-noise for a desired bit error rate is greatly reduced through the use of forward error correction techniques. Even greater coding gain is provided through the use of a concatenated coding scheme. Interleaving/deinterleaving is necessary to randomize burst errors which may appear at the input of the RS decoder. The burst correction capability length is increased in proportion to the interleave depth. The modular nature of the simulator allows for inclusion or exclusion of modules as needed. This paper describes the development and operation of the simulator, the verification of a C-language Reed-Solomon code, and the possibility of using Comdisco SPW(tm) as a tool for determining optimal error control schemes.

Noble, Viveca K.↗

PPM demodulation for Reed-Solomon decoding for the optical space channel

The use of Reed-Solomon (RS) block codes over the pulse position modulated (PPM) frames to obtain the largest degree of error correction is considered. Since RS codes can correct both symbol errors and symbol erasures, a question arises as to the best way to demodulate the PPM laser fields in order to generate the input symbols for the RS decoder. The method selected for demodulating (converting the received laser field to digital symbols) defines the erasure and transmitted symbols of the laser link, and therefore determines the work error probabilities of the system. Several demodulating schemes are considered, and the effect of each on RS decoding performance computed. This computation was carried out for various optical receiver models. It is shown that simple threshold decisioning of pulse slots produces performance that degrades as the background noise increases. This is caused by the generation of too many erasures for the RS decoder to handle. A decision scheme, delta-max demodulation which offers improvement over threshold decisioning by redefining the generation of an erasure is proposed.

Divsalar, D.↗

Evaluation of Open-Source Hard Real Time Software Packages

Reliable software is, at times, hard to find. No piece of software can be guaranteed to work in every situation that may arise during its use here at Glenn Research Center or in space. The job of the Software Assurance (SA) group in the Risk Management Office is to rigorously test the software in an effort to ensure it matches the contract specifications. In some cases the SA team also researches new alternatives for selected software packages. This testing and research is an integral part of the department of Safety and Mission Assurance. Real Time operation in reference to a computer system is a particular style of handing the timing and manner with which inputs and outputs are handled. A real time system executes these commands and appropriate processing within a defined timing constraint. Within this definition there are two other classifications of real time systems: hard and soft. A soft real time system is one in which if the particular timing constraints are not rigidly met there will be no critical results. On the other hand, a hard real time system is one in which if the timing constraints are not met the results could be catastrophic. An example of a soft real time system is a DVD decoder. If the particular piece of data from the input is not decoded and displayed to the screen at exactly the correct moment nothing critical will become of it, the user may not even notice it. However, a hard real time system is needed to control the timing of fuel injections or steering on the Space Shuttle; a delay of even a fraction of a second could be catastrophic in such a complex system. The current real time system employed by most NASA projects is Wind River's VxWorks operating system. This is a proprietary operating system that can be configured to work with many of NASA s needs and it provides very accurate and reliable hard real time performance. The down side is that since it is a proprietary operating system it is also costly to implement. The prospect of replacing this somewhat costly implementation is the focus of one of the SA group s current research projects. The explosion of open source software in the last ten years has led to the development of a multitude of software solutions which were once only produced by major corporations. The benefits of these open projects include faster release and bug patching cycles as well as inexpensive if not free software solutions. The main packages for hard real time solutions under Linux are Real Time Application Interface (RTAI) and two varieties of Real Time Linux (RTL), RTLFree and RTLPro. During my time here at NASA I have been testing various hard real time solutions operating as layers on the Linux Operating System. All testing is being run on an Intel SBC 2590 which is a common embedded hardware platform. The test plan was provided to me by the Software Assurance group at the start of my internship and my job has been to test the systems by developing and executing the test cases on the hardware. These tests are constructed so that the Software Assurance group can get hard test data for a comparison between the open source and proprietary implementations of hard real time solutions.

Mattei, Nicholas S.↗

Trellises and Trellis-Based Decoding Algorithms for Linear Block Codes: The Map and Related Decoding Algirithms - Part 3

In a coded communication system with equiprobable signaling, MLD minimizes the word error probability and delivers the most likely codeword associated with the corresponding received sequence. This decoding has two drawbacks. First, minimization of the word error probability is not equivalent to minimization of the bit error probability. Therefore, MLD becomes suboptimum with respect to the bit error probability. Second, MLD delivers a hard-decision estimate of the received sequence, so that information is lost between the input and output of the ML decoder. This information is important in coded schemes where the decoded sequence is further processed, such as concatenated coding schemes, multi-stage and iterative decoding schemes. In this chapter, we first present a decoding algorithm which both minimizes bit error probability, and provides the corresponding soft information at the output of the decoder. This algorithm is referred to as the MAP (maximum aposteriori probability) decoding algorithm.

Lin, Shu↗

Architecture for time or transform domain decoding of reed-solomon codes

Two pipeline (255,233) RS decoders, one a time domain decoder and the other a transform domain decoder, use the same first part to develop an errata locator polynomial .tau.(x), and an errata evaluator polynominal A(x). Both the time domain decoder and transform domain decoder have a modified GCD that uses an input multiplexer and an output demultiplexer to reduce the number of GCD cells required. The time domain decoder uses a Chien search and polynomial evaluator on the GCD outputs .tau.(x) and A(x), for the final decoding steps, while the transform domain decoder uses a transform error pattern algorithm operating on .tau.(x) and the initial syndrome computation S(x), followed by an inverse transform algorithm in sequence for the final decoding steps prior to adding the received RS coded message to produce a decoded output message.

Shao, Howard M.↗

CAN-D: Controller Area Network Decoder Codebase

CAN-D codebase takes as input a set of collected CAN data logs from a vehicle, and produces a .dbc with partial or full signal definitions. Once this .dbc file is produced it can be used to translate subsequent messages from that vehicle in real time or post-drive processing. code also provides methods for many visualizations and operations on the translated data.

Verma, Kiren E↗

Circuit for high resolution decoding of multi-anode microchannel array detectors

A circuit for high resolution decoding of multi-anode microchannel array detectors consisting of input registers accepting transient inputs from the anode array; anode encoding logic circuits connected to the input registers; midpoint pipeline registers connected to the anode encoding logic circuits; and pixel decoding logic circuits connected to the midpoint pipeline registers is described. A high resolution algorithm circuit operates in parallel with the pixel decoding logic circuit and computes a high resolution least significant bit to enhance the multianode microchannel array detector's spatial resolution by halving the pixel size and doubling the number of pixels in each axis of the anode array. A multiplexer is connected to the pixel decoding logic circuit and allows a user selectable pixel address output according to the actual multi-anode microchannel array detector anode array size. An output register concatenates the high resolution least significant bit onto the standard ten bit pixel address location to provide an eleven bit pixel address, and also stores the full eleven bit pixel address. A timing and control state machine is connected to the input registers, the anode encoding logic circuits, and the output register for managing the overall operation of the circuit.

Kasle, David B.↗

Assessing Tropical Pacific-induced Predictability of Southern California Precipitation Using a Novel Multi-input Multi-output Autoencoder

We construct a novel Multi-Input Multi-Output Autoencoder-decoder (MIMO-AE) to capture the non-linear relationship of Southern California precipitation and tropical Pacific Ocean sea surface temperature. The MIMO-AE is trained on both monthly TP-SST and SC-PRECIP anomalies simultaneously. The co-variability of the two fields in the MIMO-AE shared nonlinear latent space can be condensed into an index, termed the MIMO-AE index. We use a transfer learning approach to train a MIMO-AE on the combined dataset of 100 years of output from a historical simulation with the Energy Exascale Earth Systems Model version 1 and a segment of observational data. We further use Long Short-Term Memory networks to assess sub-seasonal predictability of SC-PRECIP using the MIMO-AE index. We find that the MIMO-AE index provides enhanced predictability of SC-PRECIP for a lead-time of up-to four months as compared to Niño 3.4 index and the El Niño Southern Oscillation Longitudinal Index.

54 ENVIRONMENTAL SCIENCES↗

A comparison of model-based VQ compression with other VQ approaches

In our previous work on Model-Based Vector Quantization (MVQ), we presented some performance comparisons (both rate distortion and decompression time) with VQ and JPED/DCT. In this paper, we compare the MVQ's rate distortion performance with Mean Removed Vector Quantization (MRVQ) and include our previous comparison with VQ. Both MVQ and MRVQ compute the mean of each vector (raster-scanned image block) and produce mean removed residual vectors by subtracting the block mean from the elements of each vector. In the case of MRVQ, a codebook of residual vectors is generated using a training set. For MVQ, an internal codebook is generated based on the statistical properties of the residual vectors, and upon correlations derived from a Human Visual System (HVS) model. In both MVQ and MRVQ, the block mean and address of the codevector from the codebook that most closely matches each input vector are transmitted to the decoder. MVQ, a single additional statistical parameter is transmitted to the decoder. For MRVQ, we assume that the codebook of residual vectors is available to the decoder. In our experiments, we found that the rate of distortion performance of MVQ is almost always better than VQ, and is comparable to MRVQ. Further, MVQ is much easier to use than either VQ or MRVQ, since the training and management of codebooks is not required.

Manohar, Mareboyana↗

Entropy agnostic data encoding and decoding

Entropy agnostic data encoding includes: receiving, by an encoder, input data including a bit string; generating a plurality of candidate codewords, including encoding the input data bit string with a plurality of binary vectors, wherein the plurality of binary vectors includes a set of deterministic biased binary vectors and a set of random binary vectors; selecting, in dependence upon a predefined criteria, one of the plurality of candidate codewords; and transmitting the selected candidate codeword to a decoder.

Seyedzadehdelcheh, Seyedmohammad↗

Entropy agnostic data encoding and decoding

Entropy agnostic data encoding includes: receiving, by an encoder, input data including a bit string; generating a plurality of candidate codewords, including encoding the input data bit string with a plurality of binary vectors, wherein the plurality of binary vectors includes a set of deterministic biased binary vectors and a set of random binary vectors; selecting, in dependence upon a predefined criteria, one of the plurality of candidate codewords; and transmitting the selected candidate codeword to a decoder.

SeyedzadehDelcheh, SeyedMohammad↗

Generative attribute optimization

A generative attribute optimization (“GAO”) system facilitates understanding of effects of changes of attribute values of an object on a characteristic of the object and automatically identifying attribute values to achieve a desired result for the characteristic. The GAO system trains a generator (encoder and decoder) using an attribute generative adversarial network. The GAO model includes the trained generator and a separately trained predictor model. The GAO model inputs an input image and modified attribute values and employs the encoder and the decoder to generate a modified image that is the input image modified based on the modified attribute values. The GAO model then employs the predictor model to that inputs the modified image and generate a prediction of a characteristic of the modified image. The GAO system may employ an optimizer to modify the attribute values until an objective based on the desired result is achieved.

97 MATHEMATICS AND COMPUTING↗

HydraGNN_Predictive_GFM_2026 - Ensemble of predictive graph foundation models for atomistic materials modeling

This release contains data and parameters of HydraGNN-based graph foundation models trained as a result of the work published in the pre-print "Exascale Multi-Task Graph Foundation Models for Imbalanced, Multi-Fidelity Atomistic Data" by M. Lupo Pasini et al. (https://arxiv.org/abs/2604.15380). We jointly train on 16 open first-principles datasets (544+ million structures covering 85+ elements) using a multi-task architecture with per-dataset heads and a scalable ADIOS2/DDStore data pipeline. On Frontier, we execute six large-scale DeepHyper hyperparameter optimization campaigns in FP64 and promote the top-performing message-passing models to sustained 2,048-node training, yielding a PaiNN-based lead model. The version of HydraGNN used to generate the outputs provided in this release is HydraGNN v5.0 (https://github.com/ORNL/HydraGNN/releases/tag/v5.0) The list of datasets used for the training of the graph foundation model is the following: 1) Alexandria [1] 2) ANI1x [2] 3) MPTrj [3] 4) Open Catalyst 2020 (OC20) [4] 5) Open Catalyst 2022 (OC22) [5] 6) Open Catalyst 2025 (OC25) [6] 7) Open Direct ir Capture 2023 (ODAC23) [7] 8) Open Materials 2024 (OMat24) [8] 9) Open Molecules 2025 (OMol25) [9] 10) OMol25-neutral (subset of OMol25 that contains only molecules with zero total charge) 11) OMol25-non-neutral (subset of OMol25 that contains only molecules with non-zero total charge) 12) Open Polymers 2026 (OPoly2026) [10] 13) Nabla2DFT [11] 14) QCML [12] 15) QM7X [reference 13] 16) transition1x [14] Dataset references: [1] J. Schmidt et al., “A dataset of 175k stable and metastable materials calculated with the PBEsol and SCAN functionals,” Scientific Data, vol. 9, p. 64, 2022. [2] J. S. Smith et al., “The ANI-1ccx and ANI-1x data sets, coupled-cluster and density functional theory properties for molecules,” Scientific Data, vol. 7, p. 134, 2020. [Online]. Available: https: //www.nature.com/articles/s41597-020-0473-z [3] A. Jain et al., “Commentary: The Materials Project: A materials genome approach to accelerating materials innovation,” APL Materials, vol. 1, no. 1, p. 011002, 07 2013. [Online]. Available: https://doi.org/10.1063/1.4812323 [4] L. Chanussot et al., “Open catalyst 2020 (oc20) dataset and community challenges,” ACS Catalysis, vol. 11, no. 10, pp. 6059–6072, 2021. [Online]. Available: https://doi.org/10.1021/acscatal.0c04525 [5] K. Tran et al., “Open catalyst 2022 (oc22) dataset and challenges for oxidation electrocatalysts,” ACS Catalysis, vol. 13, no. 5, pp. 3066–3084, 2023. [Online]. Available: https://doi.org/10.1021/acscatal.2c05426 [6] S. J. Sahoo et al., “The open catalyst 2025 (oc25) dataset and models for solid-liquid interfaces,” arXiv preprint arXiv:2509.17862, 2025. [Online]. Available: https://arxiv.org/abs/2509.17862 [7] A. Sriram et al., “The open DAC 2023 dataset and challenges for sorbent discovery in direct air capture,” ACS Central Science, vol. 10, no. 5, pp. 923–941, 2024. [8] L. Barroso-Luque et al., “Open materials 2024 (omat24) inorganic materials dataset and models,” 2024. [Online]. Available: https://arxiv.org/abs/2410.12771 [9] D. S. Levine et al., “The open molecules 2025 (OMol25) dataset, evaluations, and models,” 2025. [Online]. Available: https://arxiv.org/abs/2505.08762 [10] D. S. Levine et al., The open polymers 2026 (OPoly26) dataset and evaluations,” arXiv preprint arXiv:2512.23117, 2025. [Online]. Available: https://arxiv.org/abs/2512.23117 [11] K. Khrabrov et al., “Nabla2dft: A universal quantum chemistry dataset of drug-like molecules and a benchmark for neural network potentials,” in NeurIPS 2024 Datasets and Benchmarks Track, 2024. [Online]. Available: https://openreview.net/forum?id=ElUrNM9U8c [12] S. Ganscha et al., “The QCML dataset, quantum chemistry reference data from 33.5M DFT and 14.7B semi-empirical calculations,” Scientific Data, vol. 12, p. 406, 2025. [13] J. Hoja et al., “QM7-X, a comprehensive dataset of quantum-mechanical properties spanning the chemical space of small organic molecules,” Scientific Data, vol. 8, p. 43, 2021. [Online]. Available: https://www.nature.com/articles/s41597-021-00812-2 [14] M. Schreiner et al., “Transition1x - a dataset for building generalizable reactive machine learning potentials,” Scientific Data, vol. 9, p. 779, 2022. The folder "datasets_ADIOS2_format" contains the set of pre-processed datasets in Adaptable I/O System (ADIOS) format (https://www.exascaleproject.org/research-project/adios/) that have been used for the development and training of GFMs in this work. The "datasets_ADIOS2_format" directory contains 2 sub-directories, one for the version "v1" of the datasets and one for the version "v2" of the datasets. The version "v1" of the datasets provides values of the total energy as they are extracted from the original data as it was released by the respective institutions. The version "v2" of the datasets provides values of the energy that have been realigned. The realignment was performed by training a linear regression model that predicts the total energy as a function of the chemical composition of the atomistic structure, and then subtract such prediction from the original value of the total energy. Both folders "v1" and "v2" contain 16 sub-directories, each corresponding to an ADIOS2-formatted dataset The folder "DeepHyper-results" contains the configurational files and model's parameters for all the 186 HPO trials that were successfully completed by the scalable hyperparameter optimization (HPO) runs on Frontier. The content of the folder "DeepHyper-results" I structured as follows: 1) task-list.txt: list of mpnn name, jobid, and deephyper task id 2) gfm_${MPNN}_${JOBID}_0.${TASKID}: run directory with checkpoint files 3) gfm_${MPNN}: deephyper summary directory (*.csv) for each specific MPNN type 4) deephyper-experiment-${JOBID}: output and error logs for each job The file "deephyper-sorted.csv" contains the details of each HydraGNN model built and tested by HPO, obtained by merging the (*.csv) filed from each HPO run executed. Out of all the HPO trials, we selected 10 to continue the training of the respective HydraGNN models. Due to limited computational budget available in the LRN070 allocation we could not complete the training till convergence for all these 10 selected models. The folder "models" contains multiple sub-folders, one per each HydraGNN model trained. Each model sub-folder contains the parameters of each HydraGNN model, with multiple checkpoint-restarts. The list of sub-folders are as follows: 1) multidataset_hpo-BEST1-fp64 2) multidataset_hpo-BEST2-fp64 3) multidataset_hpo-BEST3-fp64 4) multidataset_hpo-BEST4-fp64 5) multidataset_hpo-BEST5-fp64 6) multidataset_hpo-BEST6-fp64 7) multidataset_hpo-BEST7-fp64 8) multidataset_hpo-BEST8-fp64 9) multidataset_hpo-BEST9-fp64 10) multidataset_hpo-BEST10-fp64 Within each one of these folders, additional auxiliary log files are provided with descriptions about how the training proceeded. The lead PaiNN-model is contained inside "multidataset_hpo-BEST6-fp64". The file "mlp_branch_weights" contains the parameters of the multi-layer perceptron (MLP) used to reconcile the predictions of the 16 output decoding heads of the HydragNN architectures. The MLP takes in input the chemical composition of the atomistic structure and predicts averaging weights to linearly mix the predictions of each output decoding head toward consolidating them into a single one. The folder "1.1billion-structure-inference" contains 1.1 billion atomistic structures randomly generated. Each structures is associated with energy and forces predicted with the lead-PaiNN model combined with the MLP model for reconciliation of the multi-branch predictions generated by the 16 output decoding heads. The folder "1.1billion-structure-inference" contains 9,300 (*.tar.gz) subdirectories, one per Frontier compute node used to execute the inference at exascale. Once uncompressed, each (*.tar.gz) subdirectory contains an ADIOS2 (*.bp) file container, where each atomistic structure is stored as a PyTorch-Geometric Data object. The file "export_dataset_environment_variables.sh" contains the environment variables that need to be set before running the HydraGNN code to reproduce the results provided in this dataset release. The code that can be used to load the ADIOS2 files, load HydraGNN models, and run inference is available at: https://github.com/ORNL/HydraGNN/releases/tag/v5.0

36 MATERIALS SCIENCE↗

Adaptive neural network management system

A method and computer system for managing a neural network. Data is sent into an input layer in a portion of layers of nodes in the neural network. The data moves on an encode path through the portion such that an output layer in the portion outputs encoded data. The encoded data is sent into the output layer on a decode path through the portion back to the input layer to obtain a reconstruction of the data by the input layer. A determination is made as to whether an undesired amount of error has occurred in the output layer based on the data sent into the input layer and the reconstruction of the data. A number of new nodes is added to the output layer when a determination is present that the undesired amount of the error occurred, enabling reducing the error using the number of the new nodes.

Draelos, Timothy J.↗

Serial-Turbo-Trellis-Coded Modulation with Rate-1 Inner Code

Serially concatenated turbo codes have been proposed to satisfy requirements for low bit- and word-error rates and for low (in comparison with related previous codes) complexity of coding and decoding algorithms and thus low complexity of coding and decoding circuitry. These codes are applicable to such high-level modulations as octonary phase-shift keying (8PSK) and 16-state quadrature amplitude modulation (16QAM); the signal product obtained by applying one of these codes to one of these modulations is denoted, generally, as serially concatenated trellis-coded modulation (SCTCM). These codes could be particularly beneficial for communication systems that must be designed and operated subject to limitations on bandwidth and power. Some background information is prerequisite to a meaningful summary of this development. Trellis-coded modulation (TCM) is now a well-established technique in digital communications. A turbo code combines binary component codes (which typically include trellis codes) with interleaving. A turbo code of the type that has been studied prior to this development is composed of parallel concatenated convolutional codes (PCCCs) implemented by two or more constituent systematic encoders joined through one or more interleavers. The input information bits feed the first encoder and, after having been scrambled by the interleaver, enter the second encoder. A code word of a parallel concatenated code consists of the input bits to the first encoder followed by the parity check bits of both encoders. The suboptimal iterative decoding structure for such a code is modular, and consists of a set of concatenated decoding modules one for each constituent code connected through an interleaver identical to the one in the encoder side. Each decoder performs weighted soft decoding of the input sequence. PCCCs yield very large coding gains at the cost of a reduction in the data rate and/or an increase in bandwidth.

Divsalar, Dariush↗

Transformer Neural Networks with Spatiotemporal Attention for Predictive Control and Optimization of Industrial Processes

In the context of real-time optimization and model predictive control of industrial systems, machine learning, and neural networks represent cutting-edge tools that hold promise for enhancing dynamic modeling. This work presents a novel transformer neural network architecture for real-time optimization and model predictive control. This network design includes a modified attention mechanism inspired by positional embedding attention from vision transformers and task-specific modifications to the input-output structure of the transformer’s decoder stack. Experiments were conducted using data from a 450 MW coal-fired power plant to evaluate this approach's effectiveness. The transformer neural network was compared with conventional recurrent models, including GRU and LSTM. The transformer exhibited a 6% increase in the R-squared (R2) value of predictions and an 83% reduction in mean squared error (MSE). Computation time was also reduced by 84% compared to conventional recurrent models.

Gallup, Ethan R.↗