Expressive dynamics models with nonlinear injective readouts enable reliable recovery of latent features from neural activity.

The advent of large-scale neural recordings has enabled new approaches that aim to discover the computational mechanisms of neural circuits by understanding the rules that govern how their state evolves over time. While these neural dynamics cannot be directly measured, they can typically be approximated by low-dimensional models in a latent space. How these models represent the mapping from latent space to neural space can affect the interpretability of the latent representation. We show that typical choices for this mapping (e.g., linear or MLP) often lack the property of injectivity, meaning that changes in latent state are not obligated to affect activity in the neural space. During training, non-injective readouts incentivize the invention of dynamics that misrepresent the underlying system and the computation it performs. Combining our injective Flow readout with prior work on interpretable latent dynamics models, we created the Ordinary Differential equations autoencoder with Injective Nonlinear readout (ODIN), which learns to capture latent dynamical systems that are nonlinearly embedded into observed neural activity via an approximately injective nonlinear mapping. We show that ODIN can recover nonlinearly embedded systems from simulated neural activity, even when the nature of the system and embedding are unknown. Additionally, we show that ODIN enables the unsupervised recovery of underlying dynamical features (e.g., fixed points) and embedding geometry. When applied to biological neural recordings, ODIN can reconstruct neural activity with comparable accuracy to previous state-of-the-art methods while using substantially fewer latent dimensions. Overall, ODIN's accuracy in recovering ground-truth latent features and ability to accurately reconstruct neural activity with low dimensionality make it a promising method for distilling interpretable dynamics that can help explain neural computation.

Expressive architectures enhance interpretability of dynamics-based neural population models.

Artificial neural networks that can recover latent dynamics from recorded neural activity may provide a powerful avenue for identifying and interpreting the dynamical motifs underlying biological computation. Given that neural variance alone does not uniquely determine a latent dynamical system, interpretable architectures should prioritize accurate and low-dimensional latent dynamics. In this work, we evaluated the performance of sequential autoencoders (SAEs) in recovering latent chaotic attractors from simulated neural datasets. We found that SAEs with widely-used recurrent neural network (RNN)-based dynamics were unable to infer accurate firing rates at the true latent state dimensionality, and that larger RNNs relied upon dynamical features not present in the data. On the other hand, SAEs with neural ordinary differential equation (NODE)-based dynamics inferred accurate rates at the true latent state dimensionality, while also recovering latent trajectories and fixed point structure. Ablations reveal that this is mainly because NODEs (1) allow use of higher-capacity multi-layer perceptrons (MLPs) to model the vector field and (2) predict the derivative rather than the next state. Decoupling the capacity of the dynamics model from its latent dimensionality enables NODEs to learn the requisite low-D dynamics where RNN cells fail. Additionally, the fact that the NODE predicts derivatives imposes a useful autoregressive prior on the latent states. The suboptimal interpretability of widely-used RNN-based dynamics may motivate substitution for alternative architectures, such as NODE, that enable learning of accurate dynamics in low-dimensional latent spaces.

A large-scale neural network training framework for generalized estimation of single-trial population dynamics.

Achieving state-of-the-art performance with deep neural population dynamics models requires extensive hyperparameter tuning for each dataset. AutoLFADS is a model-tuning framework that automatically produces high-performing autoencoding models on data from a variety of brain areas and tasks, without behavioral or task information. We demonstrate its broad applicability on several rhesus macaque datasets: from motor cortex during free-paced reaching, somatosensory cortex during reaching with perturbations, and dorsomedial frontal cortex during a cognitive timing task.

SemNet: Using Local Features to Navigate the Biomedical Concept Graph.

Literature-Based Discovery (LBD) aims to connect scientists across silos by assembling models of the literature to reveal previously hidden connections. Unfortunately, LBD systems have been unable to achieve user adoption on a large scale. This work develops opens source software in Python to convert a database of semantic predications of all of PubMed's 27.9 million indexed abstracts into a semantic inference network and biomedical concept graph in Neo4j. The developed software, called SemNet, queries a modified version of the publicly available SemMedDB and computes feature vectors on source-target pairs. Each unique United Medical Language System (UMLS) concept is represented as a node and each predication as an edge. Each node is assigned one of 132 node labels (e.g., Amino Acid, Peptide, or Protein (AAPP); Gene or Genome (GG); etc.) and each edge is labeled with one of 58 predications (e.g. treats, causes, inhibits, etc.). SemNet computes a single feature value for each metapath, or sequence of node types, between a source node and user-specified target node(s). Several different types of metapath-based features (count, degree weighted path count, and HeteSim metric) are computed and vectorized. SemNet employs an unsupervised learning algorithm for rank aggregation (ULARA) to rank identified source nodes that are most relevant to the user-specified target nodes(s). Statistical analysis of correlation among identified source nodes or resultant literature network features are used to identify patterns that can guide future research. Analysis of high residual nodes is used to compare and contrast SemNet rankings between different targets of interest. An example SemNet use case is presented to assess "the differential impact of smoking on cognition in males and females" using the following target nodes: nicotine, learning, memory, tetrahydrocannabinol (THC), cigarette smoke, X chromosome, and Y chromosome. Detailed rankings are discussed. Overall results suggest a hypothesis where smoking negatively impacts cognition to a greater extent in females, but smoking has stronger cardiovascular impacts in males. In summary, SemNet provides an adoptable method for efficient LBD of PubMed that extends beyond omics-only relationships to true multi-scalar connections that can provide actionable insight for predictive medicine, research prioritization, and clinical care.