Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userLight curve classification with recurrent neural networks for GOTO: dealing with imbalanced data
84
0
0.0
(
0
)
Ask ChatGPT about the research
No Arabic abstract
The advent of wide-field sky surveys has led to the growth of transient and variable source discoveries. The data deluge produced by these surveys has necessitated the use of machine learning (ML) and deep learning (DL) algorithms to sift through the vast incoming data stream. A problem that arises in real-world applications of learning algorithms for classification is imbalanced data, where a class of objects within the data is underrepresented, leading to a bias for over-represented classes in the ML and DL classifiers. We present a recurrent neural network (RNN) classifier that takes in photometric time-series data and additional contextual information (such as distance to nearby galaxies and on-sky position) to produce real-time classification of objects observed by the Gravitational-wave Optical Transient Observer (GOTO), and use an algorithm-level approach for handling imbalance with a focal loss function. The classifier is able to achieve an Area Under the Curve (AUC) score of 0.972 when using all available photometric observations to classify variable stars, supernovae, and active galactic nuclei. The RNN architecture allows us to classify incomplete light curves, and measure how performance improves as more observations are included. We also investigate the role that contextual information plays in producing reliable object classification.
rate research
Read More
Large-scale sky surveys have played a transformative role in our understanding of astrophysical transients, only made possible by increasingly powerful machine learning-based filtering to accurately sift through the vast quantities of incoming data generated. In this paper, we present a new real-bogus classifier based on a Bayesian convolutional neural network that provides nuanced, uncertainty-aware classification of transient candidates in difference imaging, and demonstrate its application to the datastream from the GOTO wide-field optical survey. Not only are candidates assigned a well-calibrated probability of being real, but also an associated confidence that can be used to prioritise human vetting efforts and inform future model optimisation via active learning. To fully realise the potential of this architecture, we present a fully-automated training set generation method which requires no human labelling, incorporating a novel data-driven augmentation method to significantly improve the recovery of faint and nuclear transient sources. We achieve competitive classification accuracy (FPR and FNR both below 1%) compared against classifiers trained with fully human-labelled datasets, whilst being significantly quicker and less labour-intensive to build. This data-driven approach is uniquely scalable to the upcoming challenges and data needs of next-generation transient surveys. We make our data generation and model training codes available to the community.
Radar pulse streams exhibit increasingly complex temporal patterns and can no longer rely on a purely value-based analysis of the pulse attributes for the purpose of emitter classification. In this paper, we employ Recurrent Neural Networks (RNNs) to efficiently model and exploit the temporal dependencies present inside pulse streams. With the purpose of enhancing the network prediction capability, we introduce two novel techniques: a per-sequence normalization, able to mine the useful temporal patterns; and attribute-specific RNN processing, capable of processing the extracted information effectively. The new techniques are evaluated with an ablation study and the proposed solution is compared to previous Deep Learning (DL) approaches. Finally, a comparative study on the robustness of the same approaches is conducted and its results are presented.
We empirically characterize the performance of discriminative and generative LSTM models for text classification. We find that although RNN-based generative models are more powerful than their bag-of-words ancestors (e.g., they account for conditional dependencies across words in a document), they have higher asymptotic error rates than discriminatively trained RNN models. However we also find that generative models approach their asymptotic error rate more rapidly than their discriminative counterparts---the same pattern that Ng & Jordan (2001) proved holds for linear classification models that make more naive conditional independence assumptions. Building on this finding, we hypothesize that RNN-based generative classification models will be more robust to shifts in the data distribution. This hypothesis is confirmed in a series of experiments in zero-shot and continual learning settings that show that generative models substantially outperform discriminative models.
We propose two deep neural network architectures for classification of arbitrary-length electrocardiogram (ECG) recordings and evaluate them on the atrial fibrillation (AF) classification data set provided by the PhysioNet/CinC Challenge 2017. The first architecture is a deep convolutional neural network (CNN) with averaging-based feature aggregation across time. The second architecture combines convolutional layers for feature extraction with long-short term memory (LSTM) layers for temporal aggregation of features. As a key ingredient of our training procedure we introduce a simple data augmentation scheme for ECG data and demonstrate its effectiveness in the AF classification task at hand. The second architecture was found to outperform the first one, obtaining an $F_1$ score of $82.1$% on the hidden challenge testing set.
We introduce a convolutional recurrent neural network (CRNN) for music tagging. CRNNs take advantage of convolutional neural networks (CNNs) for local feature extraction and recurrent neural networks for temporal summarisation of the extracted features. We compare CRNN with three CNN structures that have been used for music tagging while controlling the number of parameters with respect to their performance and training time per sample. Overall, we found that CRNNs show a strong performance with respect to the number of parameter and training time, indicating the effectiveness of its hybrid structure in music feature extraction and feature summarisation.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
