Deep Recurrent Neural Network (Deep-RNN) for Classification of Nonlinear Data

Mishra, Debasmita; Naik, Bighnaraj; Sahoo, Ronali Madhusmita; Nayak, Janmenjoy

doi:10.1007/978-981-15-2449-3_17

Cited by 10 publications

(6 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Dengan berbagai masalah yang dapat dicangkup, pendekatan arsitektur neural network turut mengalami perkembangan pada permasalahan yang spesifik. Klasifikasi dan rekognisi pola (classification and pattern recognition) pada serangkaian data (data series) merupakan salah satu pendekatan yang dimuat oleh neural network [10], [11]. Penerapan recurrent neural network (RNN) dapat dijadikan pendekatan pada permasalahan data series, menimbang kemampuan aristektur tidak hanya melihat setiap data point sebagai bagian yang terpisah, namun juga meninjau hubungan antar data point-nya [12].…”

Section: B Machine Learning Neural Networkunclassified

Evaluasi Model Machine Learning Klasifikasi Gerak Tangan Untuk Sistem Kontrol Prototipe Prostesis Tangan

Putra¹,

Fauzi²,

Prasad³

et al. 2023

JTE

View full text Add to dashboard Cite

Hambatan berupa kehilangan fungsi bagian tubuh akan menyebabkan kesulitan dalam melakukan kegiatan secara normal. Dalam penerapan sensor electromyography (EMG) dan electroencephalography (EEG) yang kurang baik dalam mengimbangi berbagai macam kondisi fisik manusia, sensor force sensing resistor (FSR) dapat menjadi alternatif pengganti EMG dan EEG pada prostesis tangan. Dalam perencanaan model neural network, data yang dibutuhkan pada actual output hanya berupa hand gesture pada orang non pasien pasca amputasi. Long Short Term Memory (LSTM) digunakan karena dapat menangani proses data dalam jangka panjang yang merupakan salah satu keadaan timbul dalam pengolahan data sekuensial.evaluasi metrics yang dihasilkan berupa nilai accuracy pada data training dengan epoch 200 dan accuracy pada data testing. Hasil pertama dengan tanpa variasi dropout menunjukkan nilai accuracy pada training 0,9449 dan accuracy pada testing 0,961 dengan nilai loss pada training 0,1284 dan loss pada testing 0,0717. Hasil kedua dengan variasi dropout menunjukkan nilai accuracy pada training 0,9699 dan accuracy pada testing 0,9688 dengan nilai loss pada training 0,0803 dan loss pada testing 0,1061. evaluasi metrics accuracy yang dihasilkan pada dataset telah melampui nilai 0,9. Hal ini mengindikasikan model telah berjalan dengan baik untuk klasifikasi pada 11 gerakan. Kata Kunci— Sistem kontrol; machine learning; gerakan tangan; prostesis tangan.

show abstract

Section: B Machine Learning Neural Networkunclassified

Evaluasi Model Machine Learning Klasifikasi Gerak Tangan Untuk Sistem Kontrol Prototipe Prostesis Tangan

Putra¹,

Fauzi²,

Prasad³

et al. 2023

JTE

View full text Add to dashboard Cite

show abstract

“…In the RNN, not only the final output but also the output of each time step is used to compute an error and the error in each time step is back-propagated to the network. This technique is called "Backpropagation Through Time" or BPTT [42], [43], [44], [45]. Furthermore, the RNN faces unique challenges with time series and sequential data.…”

Section: B Related Work and Topicsmentioning

confidence: 99%

“…In contrast, in a usual RNN, the error at each time step is computed and backpropagated to the network. This type of backpropagation in an RNN is called Backpropagation Through Time (BPTT) as explained in the introduction section [42], [43], [44]. Because our RNN works over iteration rather than time, we call this framework Backpropagation Through Iteration (BPTI) to avoid confusion in this paper.…”

Section: Backpropagation Through Iteration (Bpti)mentioning

confidence: 99%

A Deep Recurrent Neural Network with Gated Momentum Unit for CT Image Reconstruction

Ikuta¹

2021

Preprint

View full text Add to dashboard Cite

<div><div><div><p>Many algorithms and methods have been proposed for Computed Tomography (CT) image reconstruction, partic- ularly with the recent surge of interest in machine learning and deep learning methods. The majority of recently proposed methods are, however, limited to the image domain processing where deep learning is used to learn the mapping from a noisy image data set to a true image data set. While deep learning-based methods can produce higher quality images than conventional model-based post-processing algorithms, these methods have lim- itations. Deep learning-based methods used in the image domain are not sufficient for compensating for lost information during a forward and a backward projection in CT image reconstruction especially with a presence of high noise. In this paper, we propose a new Recurrent Neural Network (RNN) architecture for CT image reconstruction. We propose the Gated Momentum Unit (GMU) that has been extended from the Gated Recurrent Unit (GRU) but it is specifically designed for image processing inverse problems. This new RNN cell performs an iterative optimization with an accelerated convergence. The GMU has a few gates to regulate information flow where the gates decide to keep important long-term information and discard insignificant short- term detail. Besides, the GMU has a likelihood term and a prior term analogous to the Iterative Reconstruction (IR). This helps ensure estimated images are consistent with observation data while the prior term makes sure the likelihood term does not overfit each individual observation data. We conducted a synthetic image study along with a real CT image study to demonstrate this proposed method achieved the highest level of Peak Signal to Noise Ratio (PSNR) and Structure Similarity (SSIM). Also, we showed this algorithm converged faster than other well-known methods.</p></div></div></div>

show abstract

“…Considering the correlation between words in the text, the authors in [20] proposed the latent Dirichlet allocation model (LDA), which added the Dirichlet prior distribution to the polynomial distribution of texts, topics, and words. Sentence-based approaches, such as convolutional neural networks (CNNs) [21], [22] and recurrent neural networks (RNNs) [23], [24], presented the numerical representation of text at the sentence level, which was effectively adopted in natural language processing (NLP) [25], [26]. However, the CNN and RNN input was based on the word vector generated by word embedding methods.…”

Section: A Text Miningmentioning

confidence: 99%