Visualizing the Hidden Activity of Artificial Neural Networks

By Paulo Eduardo Rauber, Samuel G. Fadel, Alexandre Xavier Falcão, Alexandru Cristian Telea. 2017

311 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Biophysics, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Biophysics

In Computer Vision and Pattern Recognition

Citations in the corpus

Unsupervised Learning of Neural Networks to Explain Neural Networks (2018, background): “Instead of analyzing network features from a global view [36, 26, 23], Bau et al.”
Machine learning and soil sciences: a review aided by machine learning tools (2019, methodology): “Rauber et al. (2017) used the activation maps projected into a 2-”
Visual Analytics in Deep Learning: An Interrogative Survey for the Next Frontiers (2018, unknown): “Another paper studies closely how data instances are transformed as their information is passed deep through the network, which in effect visualizes how the neural network separates various classes along approximated decision boundaries [48].”
RuleMatrix: Visualizing and Understanding Classifiers with Rules (2018, background): “More recently, a number of visual analytics methods have been developed to support the analysis of complex deep neural networks [5, 20, 21, 25, 27, 30, 35, 40].”
Anti-Money Laundering in Bitcoin: Experimenting with Graph Convolutional Networks for Financial Forensics (2019, unknown): “[20].”
Interpreting Black-Box Classifiers Using Instance-Level Visual Explanations (2017, unknown): “have been developed to visualize decisions trees [22] and neural networks [14, 19].”
Manifold: A Model-Agnostic Framework for Interpretation and Diagnosis of Machine Learning Models (2018, background): “As discussed earlier, many existing visual analytics solutions focus on interpreting the internal working mechanism of a speciﬁc model type [12,21,25,32,37,41].”
Do Convolutional Neural Networks Learn Class Hierarchy? (2017, methodology): “Scatter plots have also been used to reveal similarities between neurons [13, 53] and to compare learning trajectories of multiple networks [20].”
ActiVis: Visual Exploration of Industry-Scale Deep Neural Network Models (2017, background): “[31] studied how 2-D projected view of instance activations and neuron filters can help users better understand neural network models.”
Explainable Artificial Intelligence: a Systematic Review (2020, methodology): “Similarly, Hidden Activity Visualization [218] uses t-SNE to visualize the projections of the activation values of the hidden neurons as a 2D scatter-plot with points coloured according to the class of the instances originating them.”
Artificial neural networks trained through deep reinforcement learning discover control strategies for active flow control (2018, background): “A number of properties of each individual network, such as the variations in architecture, make systematic analysis challenging (Rauber et al., 2017; Schmidhuber, 2015b).”
Under the Hood of Neural Networks: Characterizing Learned Representations by Functional Neuron Populations and Network Ablations (2020, methodology): “, 2018) have been widely used to visualize the high dimensional activation-space of neural networks to identify the role of network areas in solving a given task (Liu et al., 2016; Rauber et al., 2016; Elloumi et al., 2018; V., 2019).”
Understanding Hidden Memories of Recurrent Neural Networks (2017, methodology): “To generate an overview for hidden units or neurons, the machine learning community has widely adopted projection methods (e.g., t-SNE and PCA) to explore or illustrate the patterns inside model’s learned features [35].”
Towards better analysis of machine learning models: A visual analytics perspective (2017, background): “Point-based techniques [18, 20] reveal the relationships between neural network components, such as neurons or learned representations, by using scatter-plots.”
LSTMVis: A Tool for Visual Analysis of Hidden State Dynamics in Recurrent Neural Networks (2016, background): “[25], in that we are concerned with interpreting the hidden states of neural network models.”