An official website of the United States government
Optical network failure management (ONFM) is a promising application of machine learning (ML) to optical networking. Typical ML-based ONFM approaches exploit historical monitored data, retrieved in a specific domain (e.g., a link or a network), to train supervised ML models and learn failure characteristics (a signature) that will be helpful upon future failure occurrence in that domain. Unfortunately, in operational networks, data availability often constitutes a practical limitation to the deployment of ML-based ONFM solutions, due to scarce availability of labeled data comprehensively modeling all possible failure types. One could purposely inject failures to collect training data, but this is time consuming and not desirable by operators. A possible solution is transfer learning (TL), i.e., training ML models on a source domain (SD), e.g., a laboratory testbed, and then deploying trained models on a target domain (TD), e.g., an operator network, possibly fine-tuning the learned models by re-training with few TD data. Moreover, in those cases when TL re-training is not successful (e.g., due to the intrinsic difference in SD and TD), another solution is domain adaptation, which consists of combining unlabeled SD and TD data before model training. We investigate domain adaptation and TL for failure detection and failure-cause identification across different lightpaths leveraging real optical SNR data. We find that for the considered scenarios, up to 20% points of accuracy increase can be obtained with domain adaptation for failure detection, while for failure-cause identification, only combining domain adaptation with model re-training provides significant benefit, reaching 4%–5% points of accuracy increase in the considered cases.
more »
« less
Applying Transfer Learning for Street-Scale Nuisance Flood Forecasting in Coastal-Urban Cities
An important challenge with Machine Learning (ML) is itstransferability; i.e., whether a ML model trained on one set of data canbe applied to a second set of data without requiring full re-training ofthe model. Transfer Learning (TL) addresses this challenge bytransferring knowledge learned in the source domain (the data it wastrained on) to the target domain (a second set of data that isstatistically different but related, which the model was not trainedon). This study investigates the use of TL for street-scale nuisanceflood forecasting by exploring whether a ML model trained for one set ofstreets can effectively forecast flooding for another set of streets inthe same city. TL is explored using a Long Short-Term Memory (LSTM)model trained on data for the flood-prone streets of Norfolk City,Virginia. The results show that full-weight re-training proved mosteffective and minimal re-training of only the output layer wasinsufficient. The advantage of TL was most pronounced when target datawas limited, meaning data collected at the new water depth sensorlocation included generally less than 18 flood events. As target dataincreased beyond 18 flood events, the benefit of TL diminished relativeto training a ML model directly on the local flood events. Thesefindings can assist cities as they implement street-scale flood sensingsystems to create accurate forecasts for new sensing locations that donot yet have sufficient data records to train a local ML model.
more »
« less
- Award ID(s):
- 2209139
- PAR ID:
- 10699028
- Publisher / Repository:
- ESS Open Archive
- Date Published:
- Format(s):
- Medium: X
- Institution:
- University of Virginia
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
null (Ed.)Given its demonstrated ability in analyzing and revealing patterns underlying data, Deep Learning (DL) has been increasingly investigated to complement physics-based models in various aspects of smart manufacturing, such as machine condition monitoring and fault diagnosis, complex manufacturing process modeling, and quality inspection. However, successful implementation of DL techniques relies greatly on the amount, variety, and veracity of data for robust network training. Also, the distributions of data used for network training and application should be identical to avoid the internal covariance shift problem that reduces the network performance applicability. As a promising solution to address these challenges, Transfer Learning (TL) enables DL networks trained on a source domain and task to be applied to a separate target domain and task. This paper presents a domain adversarial TL approach, based upon the concepts of generative adversarial networks. In this method, the optimizer seeks to minimize the loss (i.e., regression or classification accuracy) across the labeled training examples from the source domain while maximizing the loss of the domain classifier across the source and target data sets (i.e., maximizing the similarity of source and target features). The developed domain adversarial TL method has been implemented on a 1-D CNN backbone network and evaluated for prediction of tool wear propagation, using NASA's milling dataset. Performance has been compared to other TL techniques, and the results indicate that domain adversarial TL can successfully allow DL models trained on certain scenarios to be applied to new target tasks.more » « less
-
Compound flood risk assessments require probabilistic estimates of flood depths and extents that are derived from compound flood models. It is essential to simulate a wide range of flood driver conditions to capture the full range of variability in resultant flooding. Although recent advancements in computational resources and the development of faster compound flood models allow for more rapid simulations, generating a large enough set of storm events for boundary conditions remains a challenge. In this study, we introduce a statistical framework designed to generate many synthetic but physically plausible compound events, including storm-tide hydrographs and rainfall fields, which can serve as boundary conditions for dynamic compound flood models. We apply the proposed framework to Gloucester City in New Jersey, as a case study. The results demonstrate its effectiveness in producing synthetic events covering the unobserved regions of the parameter space. We use flood model simulations to assess the importance of explicitly accounting for variability in mean sea level (m.s.l.) and tides in generating the boundary conditions. Results highlight that m.s.l. anomalies and tidal conditions alone can lead to differences in flood depths exceeding 1 and 1.2 m, respectively, in parts of Gloucester City. While we use historically observed events, the framework can be applied to model output data including hindcasts or future projections.more » « less
-
Yortsos, Yannis (Ed.)Abstract Transfer learning (TL), which enables neural networks (NNs) to generalize out-of-distribution via targeted re-training, is becoming a powerful tool in scientific machine learning (ML) applications such as weather/climate prediction and turbulence modeling. Effective TL requires knowing (1) how to re-train NNs? and (2) what physics are learned during TL? Here, we present novel analyses and a framework addressing (1)–(2) for a broad range of multi-scale, nonlinear, dynamical systems. Our approach combines spectral (e.g. Fourier) analyses of such systems with spectral analyses of convolutional NNs, revealing physical connections between the systems and what the NN learns (a combination of low-, high-, band-pass filters and Gabor filters). Integrating these analyses, we introduce a general framework that identifies the best re-training procedure for a given problem based on physics and NN theory. As test case, we explain the physics of TL in subgrid-scale modeling of several setups of 2D turbulence. Furthermore, these analyses show that in these cases, the shallowest convolution layers are the best to re-train, which is consistent with our physics-guided framework but is against the common wisdom guiding TL in the ML literature. Our work provides a new avenue for optimal and explainable TL, and a step toward fully explainable NNs, for wide-ranging applications in science and engineering, such as climate change modeling.more » « less
-
Flooding occurs at different scales and unevenly affects urban populations based on the broader social, ecological, and technological system (SETS) characteristics particular to cities. As hydrological models improve in spatial scale and account for more mechanisms of flooding, there is a continuous need to examine the re- lationships between flood exposure and SETS drivers of flood vulnerability. In this study, we related fine-scale measures of future flood exposure—the First Street Foundation’s Flood Factor and estimated change in chance of extreme flood exposure—to SETS indicators like building age, poverty, and historical redlining, at the parcel and census block group (CBG) scales in Portland, OR, Phoenix, AZ, Baltimore, MD, and Atlanta, GA. We used standard regression models and accounted for spatial bias in relationships. The results show that flood exposure was more often correlated with SETS variables at the parcel scale than at the CBG scale, indicating scale dependence. However, these relationships were often inconsistent among cities, indicating place-dependence. We found that marginalized populations were significantly more exposed to future flooding at the CBG scale. Combining newly-available, high-resolution future flood risk estimates with SETS data available at multiple scales offers cities a new set of tools to assess the exposure and multi-dimensional vulnerability of populations. These tools will better equip city managers to proactively plan and implement equitable interventions to meet evolving hazard exposure.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript
- Posted Content:
- https://doi.org/10.22541/essoar.175492904.46217552/v1
