Convolutional neural networks are rapidly gaining popularity in regulatory genomics. Typically, these networks have a stack of convolutional and pooling layers, followed by one or more fully connected layers. In genomics, the same positional patterns are often present across multiple convolutional channels. Therefore, in current state-of-the-art networks, there exists significant redundancy in the representations learned by standard fully connected layers. We present a new separable fully connected layer that learns a weights tensor that is the outer product of positional weights and cross-channel weights, thereby allowing the same positional patterns to be applied across multiple convolutional channels. Decomposing positional and cross-channel weights further enables us to readily impose biologically-inspired constraints on positional weights, such as symmetry. We also propose a novel regularizer that penalizes curvature in the positional weights. Using experiments on simulated and in vivo datasets, we show that networks that incorporate our separable fully connected layer outperform conventional models with analogous architectures and the same number of parameters. Additionally, our networks are more robust to hyperparameter tuning, have more informative gradients, and produce importance scores that are more consistent with known biology than conventional deep neural networks. Availability Our implementation is available at: https://github.com/kundajelab/keras/tree/keras_1 A gist illustrating model setup is at: goo.gl/gYooaa
[1]
Avanti Shrikumar,et al.
Reverse-complement parameter sharing improves deep learning models for genomics
,
2017,
bioRxiv.
[2]
Howard Y. Chang,et al.
Lineage-specific and single cell chromatin accessibility charts human hematopoiesis and leukemia evolution
,
2016,
Nature Genetics.
[3]
Avanti Shrikumar,et al.
Learning Important Features Through Propagating Activation Differences
,
2017,
ICML.
[4]
ENCODEConsortium,et al.
An Integrated Encyclopedia of DNA Elements in the Human Genome
,
2012,
Nature.
[5]
Jimmy Ba,et al.
Adam: A Method for Stochastic Optimization
,
2014,
ICLR.
[6]
S. Ferrari,et al.
Author contributions
,
2021
.
[7]
Yoshua Bengio,et al.
Understanding the difficulty of training deep feedforward neural networks
,
2010,
AISTATS.
[8]
David R. Kelley,et al.
Basset: learning the regulatory code of the accessible genome with deep convolutional neural networks
,
2015,
bioRxiv.
[9]
B. Frey,et al.
Predicting the sequence specificities of DNA- and RNA-binding proteins by deep learning
,
2015,
Nature Biotechnology.