Filters








137,880 Hits in 5.0 sec

Classical Structured Prediction Losses for Sequence to Sequence Learning [article]

Sergey Edunov, Myle Ott, Michael Auli, David Grangier, Marc'Aurelio Ranzato
2018 arXiv   pre-print
In this paper, we survey a range of classical objective functions that have been widely used to train linear models for structured prediction and apply them to neural sequence to sequence models.  ...  Our experiments show that these losses can perform surprisingly well by slightly outperforming beam search optimization in a like for like setup.  ...  Conclusion We present a comprehensive comparison of classical losses for structured prediction and apply them to a strong neural sequence to sequence model.  ... 
arXiv:1711.04956v5 fatcat:sblsldsdcvav3f5rp362fse6wa

Classical Structured Prediction Losses for Sequence to Sequence Learning

Sergey Edunov, Myle Ott, Michael Auli, David Grangier, Marc'Aurelio Ranzato
2018 Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers)  
In this paper, we survey a range of classical objective functions that have been widely used to train linear models for structured prediction and apply them to neural sequence to sequence models.  ...  Our experiments show that these losses can perform surprisingly well by slightly outperforming beam search optimization in a like for like setup.  ...  Conclusion We present a comprehensive comparison of classical losses for structured prediction and apply them to a strong neural sequence to sequence model.  ... 
doi:10.18653/v1/n18-1033 dblp:conf/naacl/EdunovOAGR18 fatcat:b6csjqyzjze3dae4vnga7b3b2e

Quantum Deep Learning for Mutant COVID-19 Strain Prediction [article]

Yu-Xin Jin, Jun-Jie Hu, Qi Li, Zhi-Cheng Luo, Fang-Yan Zhang, Hao Tang, Kun Qian, Xian-Min Jin
2022 arXiv   pre-print
Early prediction of possible variants (especially spike protein) of COVID-19 epidemic strains based on available mutated SARS-CoV-2 RNA sequences may lead to early prevention and treatment.  ...  At last, evidences that quantum-inspired algorithms promote the classical deep learning and hybrid models effectively predict the mutant strains are strong.  ...  For classical model, we add a penalty term to discriminator loss just for successfully training.  ... 
arXiv:2203.03556v1 fatcat:3wmo3inpkbf2lmvwoymvirwwhu

Car-Traffic Forecasting: A Representation Learning Approach

Ali Ziat, Gabriella Contardo, Nicolas Baskiotis, Ludovic Denoyer
2015 International Conference on Machine Learning  
The model has been tested for a concrete application: cartraffic forecasting where each time series characterizes a particular road and where the graph structure corresponds to the road map of the city  ...  We address the problem of learning over multiple inter-dependent temporal sequences where dependencies are modeled by a graph.  ...  The parameters γ will be learned to minimize the mean square error between the prediction h γ (z (t) i ) and z (t+1) i . • At last, term 3 corresponds to a structural regularity over the graph structure  ... 
dblp:conf/icml/ZiatCBD15 fatcat:oykhyp74bbgmxhstiqfr6w3pze

A Novel Dynamic Attack on Classical Ciphers Using an Attention-Based LSTM Encoder-Decoder Model

Ezat Ahmadzadeh, Hyunil Kim, Ongee Jeong, Inkyu Moon
2021 IEEE Access  
This paper presents a novel, dynamic way to attack classical ciphers by using an attention-based LSTM encoder-decoder for different ciphertext sequence lengths.  ...  Cryptanalysis using deep learning-based methods to identify weaknesses in ciphers has not been thoroughly studied.  ...  Since the ciphertext lengths are long, it would be difficult for classical LSTM networks to learn.  ... 
doi:10.1109/access.2021.3074268 fatcat:ax6e446vubcbvpvbglstnimglq

Learning Embeddings for Completion and Prediction of Relationnal Multivariate Time-Series

Ali Ziat, Gabriella Contardo, Nicolas Baskiotis, Ludovic Denoyer
2016 The European Symposium on Artificial Neural Networks  
We propose a model that is able to simultaneously fill in missing values and predict future ones.  ...  This approach is based on representation learning techniques, where temporal data are represented in a latent vector space so as to capture the dynamicity of the process and also the relations between  ...  In comparison to baselines models that have been developed for prediction only or completion only, our approach shows interesting performance and is able to simultaneously complete missing values and predict  ... 
dblp:conf/esann/ZiatCBD16 fatcat:laf5uqdc3nhlnhzkvu45vgwfei

Structured Recommendation [article]

Dawei Chen, Lexing Xie, Aditya Krishna Menon, Cheng Soon Ong
2017 arXiv   pre-print
We propose an approach to sequence recommendation based on the structured support vector machine. For prediction, we modify the inference procedure to avoid predicting loops in the sequence.  ...  Motivated by trajectory recommendation, we focus on sequential structures but in contrast to classical Viterbi decoding we require that valid predictions are sequences with no repeated elements.  ...  Classic structured prediction does not constrain the output sequence, and having such a path constraint makes both inference and learning harder.  ... 
arXiv:1706.09067v1 fatcat:e27aao47szhvlgr6u4lz6jmmq4

A Self-Supervised, Differentiable Kalman Filter for Uncertainty-Aware Visual-Inertial Odometry [article]

Brandon Wagstaff, Emmett Wise, Jonathan Kelly
2022 arXiv   pre-print
Learning-based systems have the potential to outperform classical implementations in challenging environments, but, currently, do not perform as well as classical methods in nominal settings.  ...  Herein, we introduce a framework for training a hybrid VIO system that leverages the advantages of learning and standard filtering-based state estimation.  ...  ACKNOWLEDGMENTS We are grateful to NVIDIA Corporation for providing the Quadro RTX 8000 GPU used for this research.  ... 
arXiv:2203.07207v2 fatcat:a74hcggtr5crfjogmhdalg6byi

Structured Prediction in Time Series Data

Jia Li
2017 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
Structured prediction focuses on prediction task where the outputs are structured and interdependent, contrary to the non-structured prediction which assumes that the outputs are independent of other predicted  ...  One difficulty for structured prediction is that the number of possible outputs can be exponential which makes modeling all the potential outputs intractable.  ...  Active learning for non-structured prediction has been extensively studied (Settles 2012 ). For structured prediction in time series data, there are more challenges.  ... 
doi:10.1609/aaai.v31i1.10525 fatcat:ryjbrakdlfewlbbwzeglxywdmu

Improving Protein Gamma-Turn Prediction Using Inception Capsule Networks [article]

Chao Fang, Yi Shang, Dong Xu
2018 arXiv   pre-print
Hence, it is worthwhile exploring new machine-learning methods for the prediction.  ...  One reason for the low prediction accuracy is the limited capacity of the methods; in particular, the traditional machine-learning methods like SVM may not extract high-level features well to distinguish  ...  We also like to thank Duolin Wang, Shuai Zeng, and Zhaoyu Li for their helpful discussions on Capsule Networks.  ... 
arXiv:1806.07341v1 fatcat:irubtuhpbrdgxbvdub3p4ljcsy

Deep Learning in Next-Frame Prediction: A Benchmark Review

Yufan Zhou, Haiwei Dong, Abdulmotaleb El Saddik
2020 IEEE Access  
In this paper, we introduce recent state-of-the-art next-frame prediction networks and categorize them into two architectures: sequence-to-one architecture and sequence-to-sequence architecture.  ...  INDEX TERMS Frame prediction architecture, loss function design, state-of-the-art evaluation.  ...  In image learning, the pyramid structure has been proven to be efficient for high-level feature extraction.  ... 
doi:10.1109/access.2020.2987281 fatcat:dboyrx7y35dhpifczflhnxmxe4

A Novel Prediction Method for ATP-binding Sites from Protein Primary Sequences Based on Fusion of Deep Convolutional Neural Network and Ensemble Learning

Jiazhi Song, Yanchun Liang, Guixia Liu, Rongquan Wang, Liyan Sun, Ping Zhang
2020 IEEE Access  
Previous studies often utilize classical machine learning classification algorithms to predict protein-ATP binding sites based on protein primary sequence.  ...  In this work, we introduce the deep convolutional neural network for protein-ATP binding sites prediction based on sequence information.  ...  In 2018, Hu et al proposed a hybrid prediction method called ATPbind which combines sequence-profiling and structure-based comparison for ATP binding sites prediction.  ... 
doi:10.1109/access.2020.2968847 fatcat:4hlrw27zevcmrnbdylffzlsvk4

Self-Supervised Deep Pose Corrections for Robust Visual Odometry [article]

Brandon Wagstaff, Valentin Peretroukhin, Jonathan Kelly
2020 arXiv   pre-print
Instead of regressing inter-frame pose changes directly, we build on prior work that uses data-driven learning to regress pose corrections that account for systematic errors due to violations of modelling  ...  state-of-the-art learning-only approaches.  ...  The approach described in [17] operates by learning to predict optical flow and disparity and then uses a classical RANSAC outlier rejection scheme to select a set of inlying pixels that can be used  ... 
arXiv:2002.12339v1 fatcat:wcqxk2hzazah7euhgnathnhmlm

InDiD: Instant Disorder Detection via Representation Learning [article]

Evgenia Romanenkova and Alexander Stepikin and Matvey Morozov and Alexey Zaytsev
2022 arXiv   pre-print
It approximates classic rigorous solutions but is differentiable and allows representation learning for deep models.  ...  Classic approaches for change point detection (CPD) might underperform for semi-structured sequential data because they cannot process its structure without a proper representation.  ...  for semi-structured data.  ... 
arXiv:2106.02602v3 fatcat:zor2uyfbbrglnjru7ckaehfk2y

Prediction of Protein–ATP Binding Residues Based on Ensemble of Deep Convolutional Neural Networks and LightGBM Algorithm

Jiazhi Song, Guixia Liu, Jingqing Jiang, Ping Zhang, Yanchun Liang
2021 International Journal of Molecular Sciences  
Previous studies have used classic machine-learning algorithms like support vector machine (SVM) and random forest to predict protein–ATP binding residues; however, as new machine-learning techniques are  ...  Accurately identifying protein–ATP binding residues is important for protein function annotation and drug design.  ...  In this study, PSIPRED [42] was applied to predict the secondary structure for the query residue based on the sequence information.  ... 
doi:10.3390/ijms22020939 pmid:33477866 fatcat:5zw4ulubqvfnxpvc42qndi5u2q
« Previous Showing results 1 — 15 out of 137,880 results