Unsupervised discovery of temporal sequences in high-dimensional datasets, with applications to neuroscience

@article{Mackevicius2018UnsupervisedDO,
  title={Unsupervised discovery of temporal sequences in high-dimensional datasets, with applications to neuroscience},
  author={Emily L Mackevicius and Andrew H Bahle and Alex H. Williams and Shijie Gu and Natalia I Denissenko and Mark S. Goldman and Michale S. Fee},
  journal={bioRxiv},
  year={2018}
}
The ability to identify interpretable, low-dimensional features that capture the dynamics of large-scale neural recordings is a major challenge in neuroscience. Dynamics that include repeated temporal patterns (which we call sequences), are not succinctly captured by traditional dimensionality reduction techniques such as principal components analysis (PCA) and non-negative matrix factorization (NMF). The presence of neural sequences is commonly demonstrated using visual display of trial… Expand
Discovering precise temporal patterns in large-scale neural recordings through robust and interpretable time warping
TLDR
This approach flexibly uncovers diverse dynamical firing patterns, including pulsatile responses to behavioral events, LFP-aligned oscillatory spiking, and even unanticipated patterns, like 7 Hz oscillations in rat motor cortex that are not time-locked to measured behaviors or LFP. Expand
Unsupervised clustering of temporal patterns in high-dimensional neuronal ensembles using a novel dissimilarity measure
TLDR
In an application to neural ensemble data from macaque monkey V1 cortex, SPOTDisClust can identify different moving stimulus directions on the sole basis of temporal spiking patterns and handles efficiently the additional information from increasingly large neuronal ensembles. Expand
LeMoNADe: Learned Motif and Neuronal Assembly Detection in calcium imaging videos
TLDR
This work proposes LeMoNADe, a new exploratory data analysis method that facilitates hunting for motifs in calcium imaging videos, the dominant microscopic functional imaging modality in neurophysiology. Expand
A probabilistic framework for decoding behavior from in vivo calcium imaging data
TLDR
A probabilistic framework that can be used to robustly infer behavior from calcium imaging recordings and relies on a simplified implementation of a naive Baysian classifier is described and it is shown that neuronal activity can be predicted from behavior, and that the accuracy of such reconstructions can guide the understanding of relationships that may exist between behavioral states. Expand
Greedy low-rank algorithm for spatial connectome regression
TLDR
A greedy low-rank algorithm for the connectome reconstruction problem in very high dimensions that approximates the solution by a sequence of rank-one updates which exploit the sparse and positive definite problem structure and is significantly faster than previous methods. Expand
Rank Selection in Non-negative Matrix Factorization: systematic comparison and a new MAD metric
TLDR
This work proposes a new metric for rank selection based on imputation cross-validation and systematically compares it against six other metrics while assessing the effects of data properties, showing that properties of the data heavily impact the ability of different methods. Expand
Linear dynamics of evidence integration in contextual decision making
TLDR
This study suggests that a whole space of sensory-related input signals invariantly modulates PFC responses and that decision signals emerge as the inputs are shaped by a changing circuit dynamics. Expand
Routing of Hippocampal Ripples to Subcortical Structures via the Lateral Septum
TLDR
LS is strongly activated by SPW-Rs and may convey hippocampal population events to its hypothalamic and brainstem targets, and this hypothesis was supported by optogenetic induction of HFOs in LS. Expand
Exact multiplicative updates for convolutional $\beta$-NMF in 2D
TLDR
It is shown by simulation that the updates lead to a monotonically decreasing β-divergence in terms of the mean and the standard deviation and that the corresponding convergence curves are consistent across the most common values for β. Expand
Exact Multiplicative Factor Updates for Convolutional Beta-NMF in 2D
TLDR
The convolutional NMF with the beta-divergence as cost function to two dimensions is extended and exact multiplicative updates for its factors are derived and it is proved that the cost is non-increasing under the new updates for beta between 0 and 2. Expand
...
1
2
...

References

SHOWING 1-10 OF 86 REFERENCES
Extracting spatial–temporal coherent patterns in large-scale neural recordings using dynamic mode decomposition
TLDR
The DMD approach is validated on sub-dural electrode array recordings from human subjects performing a known motor task, and the resulting analysis combines key features of performing PCA in space and power spectral analysis in time, making it particularly suitable for analyzing large-scale neural recordings. Expand
Sparse convolutional coding for neuronal ensemble identification
TLDR
Application of the proposed algorithm based on sparse convolution coding to synthetically generated datasets shows that it outperforms previous work and is able to accurately identify temporal cell ensembles even when those contain overlapping neurons or when strong background noise is present. Expand
A novel distance measure for the unsupervised clustering of temporal patterns in high-dimensional neuronal ensembles
TLDR
In an application to neural ensemble data from macaque monkey V1 cortex, SPOTDistClust can reconstruct cortical response properties in an unsupervised function, that is, without knowledge of stimulus identity or delivery times. Expand
Unsupervised clustering of temporal patterns in high-dimensional neuronal ensembles using a novel dissimilarity measure
TLDR
An unsupervised method, SPOTDisClust (Spike Pattern Optimal Transport Dissimilarity Clustering), is introduced, for their detection from high-dimensional neural ensembles, and proposes a dissimilarity measure between neuronal patterns based on optimal transport theory. Expand
Dimensionality reduction for large-scale neural recordings
TLDR
This review examines three important motivations for population studies: single-trial hypotheses requiring statistical power, hypotheses of population response structure and exploratory analyses of large data sets, and practical advice about selecting methods and interpreting their outputs. Expand
ASSET: Analysis of Sequences of Synchronous Events in Massively Parallel Spike Trains
TLDR
Improved, fully automated method which determines diagonal structures in the intersection matrix by a robust mathematical procedure, and the ability of the method to discover SSEs is demonstrated on complex data from large network simulations with embedded synfire chains. Expand
Methods for identification of spike patterns in massively parallel spike trains
TLDR
A comparative overview of methods for the detection of various types of higher-order synchronization and of spatio-temporal patterns, which combine data mining with analysis of statistical significance is given. Expand
Convex Hull Convolutive Non-Negative Matrix Factorization for Uncovering Temporal Patterns in Multivariate Time-Series Data
TLDR
The Convex Hull Convolutive Non-negative Matrix Factorization (CH-CNMF) algorithm is proposed to learn temporal patterns in multivariate time-series data and it is found that the gestures comprising the CH-C NMF basis generalize better to unseen data and capture vocal tract structure and dynamics significantly better than those comprising the CNMF-SC basis. Expand
Online Non-Negative Convolutive Pattern Learning for Speech Signals
TLDR
This paper proposes a new online implementation of CNMF and CNSC which processes input data piece-by-piece and updates learned patterns gradually with accumulated statistics and shows that, with unlimited data and computing resources, the new online learning algorithm almost surely converges to a local minimum of the objective cost function. Expand
Single-trial dynamics of motor cortex and their applications to brain-machine interfaces
TLDR
This work builds and characterized dynamical models of single-trial motor cortical activity and finds these models capture salient dynamical features of the neural population and are informative of future neural activity on single trials. Expand
...
1
2
3
4
5
...