• Corpus ID: 27862538

Visualizing LSTM decisions

@article{Westhuizen2017VisualizingLD,
  title={Visualizing LSTM decisions},
  author={Jos van der Westhuizen and Joan Lasenby},
  journal={ArXiv},
  year={2017},
  volume={abs/1705.08153}
}
Long Short-Term Memory (LSTM) recurrent neural networks are renowned for being uninterpretable "black boxes. [] Key Method The visualization techniques include input saliency by means of occlusion and derivatives, class mode visualization, and temporal outputs. Moreover, we demonstrate that LSTMs appear to extract features similar to those extracted by wavelets. It was found that deriving the inputs for saliency is a poor approximation and occlusion is a better approach. Moreover, analyzing LSTMs on…

Figures and Tables from this paper

RNNbow: Visualizing Learning via Backpropagation Gradients in Recurrent Neural Networks
TLDR
RNNbow is a web application that displays the relative gradient contributions from Recurrent Neural Network cells in a neighborhood of an element of a sequence and shows how it uncovers insights into the vanishing gradient as well as the evolution of training as the RNN works its way through a corpus.
Online anomaly detection with concept drift adaptation using recurrent neural networks
TLDR
A temporal model based on Recurrent Neural Networks for time series anomaly detection to address challenges posed by sudden or regular changes in normal behavior and demonstrates the efficacy of the proposed approach on a diverse set of synthetic, publicly available and proprietary real-world datasets.
Teaching a Machine to Diagnose a Heart Disease; Beginning from digitizing scanned ECGs to detecting the Brugada Syndrome (BrS)
TLDR
This thesis proposes a pipeline that reads in scanned images of ECGs, and transforms the encaptured signals to digital time-voltage data after several processing steps, and presents a long short-term memory (LSTM) classifier that is built based on the previously extracted data and that makes the diagnosis.

References

SHOWING 1-10 OF 34 REFERENCES
Visualizing and Understanding Recurrent Networks
TLDR
This work uses character-level language models as an interpretable testbed to provide an analysis of LSTM representations, predictions and error types, and reveals the existence of interpretable cells that keep track of long-range dependencies such as line lengths, quotes and brackets.
Visualizing and Understanding Neural Models in NLP
TLDR
Four strategies for visualizing compositionality in neural models for NLP, inspired by similar work in computer vision, including LSTM-style gates that measure information flow and gradient back-propagation, are described.
Visualizing and Understanding Convolutional Networks
TLDR
A novel visualization technique is introduced that gives insight into the function of intermediate feature layers and the operation of the classifier in large Convolutional Network models, used in a diagnostic role to find model architectures that outperform Krizhevsky et al on the ImageNet classification benchmark.
Intriguing properties of neural networks
TLDR
It is found that there is no distinction between individual highlevel units and random linear combinations of high level units, according to various methods of unit analysis, and it is suggested that it is the space, rather than the individual units, that contains of the semantic information in the high layers of neural networks.
A Critical Review of Recurrent Neural Networks for Sequence Learning
TLDR
The goal of this survey is to provide a selfcontained explication of the state of the art of recurrent neural networks together with a historical perspective and references to primary research.
Sequence to Sequence Learning with Neural Networks
TLDR
This paper presents a general end-to-end approach to sequence learning that makes minimal assumptions on the sequence structure, and finds that reversing the order of the words in all source sentences improved the LSTM's performance markedly, because doing so introduced many short term dependencies between the source and the target sentence which made the optimization problem easier.
Recurrent Batch Normalization
TLDR
It is demonstrated that it is both possible and beneficial to batch-normalize the hidden-to-hidden transition, thereby reducing internal covariate shift between time steps.
Learning to Diagnose with LSTM Recurrent Neural Networks
TLDR
This first study to empirically evaluate the ability of LSTMs to recognize patterns in multivariate time series of clinical measurements considers multilabel classification of diagnoses, and establishes the effectiveness of a simple LSTM network for modeling clinical data.
Deep Motif Dashboard: Visualizing and Understanding Genomic Sequences Using Deep Neural Networks
TLDR
A toolkit called the Deep Motif Dashboard (DeMo Dashboard) is proposed which provides a suite of visualization strategies to extract motifs, or sequence patterns from deep neural network models for TFBS classification, and results indicate that a convolutional-recurrent architecture performs the best among the three architectures.
Deep neural networks are easily fooled: High confidence predictions for unrecognizable images
TLDR
This work takes convolutional neural networks trained to perform well on either the ImageNet or MNIST datasets and finds images with evolutionary algorithms or gradient ascent that DNNs label with high confidence as belonging to each dataset class, and produces fooling images, which are then used to raise questions about the generality of DNN computer vision.
...
...