A role for cortical interneurons as adversarial discriminators.

The brain learns representations of sensory information from experience, but the algorithms by which it does so remain unknown. One popular theory formalizes representations as inferred factors in a generative model of sensory stimuli, meaning that learning must improve this generative model and inference procedure. This framework underlies many classic computational theories of sensory learning, such as Boltzmann machines, the Wake/Sleep algorithm, and a more recent proposal that the brain learns with an adversarial algorithm that compares waking and dreaming activity. However, in order for such theories to provide insights into the cellular mechanisms of sensory learning, they must be first linked to the cell types in the brain that mediate them. In this study, we examine whether a subtype of cortical interneurons might mediate sensory learning by serving as discriminators, a crucial component in an adversarial algorithm for representation learning. We describe how such interneurons would be characterized by a plasticity rule that switches from Hebbian plasticity during waking states to anti-Hebbian plasticity in dreaming states. Evaluating the computational advantages and disadvantages of this algorithm, we find that it excels at learning representations in networks with recurrent connections but scales poorly with network size. This limitation can be partially addressed if the network also oscillates between evoked activity and generative samples on faster timescales. Consequently, we propose that an adversarial algorithm with interneurons as discriminators is a plausible and testable strategy for sensory learning in biological systems.

Stability of Nanopeptides: Structure and Molecular Exchange of Self-assembled Peptide Fibers.

Often nanostructures formed by self-assembly of small molecules based on hydrophobic interactions are rather unstable, causing morphological changes or even dissolution when exposed to changes in aqueous media. In contrast, peptides offer precise control of the nanostructure through a range of molecular interactions where physical stability can be engineered in and, to a certain extent, decoupled from size via rational design. Here, we investigate a family of peptides that form beta-sheet nanofibers and demonstrate a remarkable physical stability even after attachment of poly(ethylene glycol). We employed small-angle neutron/X-ray scattering, circular dichroism spectroscopy, and molecular dynamics simulation techniques to investigate the detailed nanostructure, stability, and molecular exchange. The results for the most stable sequence did not reveal any structural alterations or unimer exchange for temperatures up to 85 °C in the biologically relevant pH range. Only under severe mechanical perturbation (i.e., tip sonication) would the fibers break up, which is reflected in a very high activation barrier for unimer exchange of ∼320 kJ/mol extracted from simulations. The results give important insight into the relation between molecular structure and stability of peptide nanostructure that is important for, e.g., biomedical applications.

Acoustic properties of Kel F-800 copolymer up to 85 GPa.

Acoustic properties of the fluorinated copolymer Kel F-800 were determined with Brillouin spectroscopy up to pressures of 85 GPa at 300 K. This research addresses outstanding issues in high-pressure polymer behavior, as to date the acoustic properties and equation of state of any polymer have not been determined above 20 GPa. We observed both longitudinal and transverse modes in all pressure domains, allowing us to calculate the C(11) and C(12) moduli, bulk, shear, and Young's moduli, and the density of Kel F-800 as a function of pressure. We found the behavior of the polymer with respect to all parameters to change drastically with pressure. As a result, we find that the data are best understood when split into two pressure regimes. At low pressures (less than ∼5 GPa), analysis of the room temperature isotherm with a semi-empirical equation of state yielded a zero-pressure bulk modulus K(o) and its derivative K(0) (') of 12.8 ± 0.8 GPa and 9.6 ± 0.7, respectively. The same analysis for the higher pressure data yielded values for K(o) and K(0) (') of 34.9 ± 1.7 GPa and 5.1 ± 0.1, respectively. We discuss this significant difference in behavior with reference to the concept of effective free volume collapse.

Efficient neural codes naturally emerge through gradient descent learning.

Human sensory systems are more sensitive to common features in the environment than uncommon features. For example, small deviations from the more frequently encountered horizontal orientations can be more easily detected than small deviations from the less frequent diagonal ones. Here we find that artificial neural networks trained to recognize objects also have patterns of sensitivity that match the statistics of features in images. To interpret these findings, we show mathematically that learning with gradient descent in neural networks preferentially creates representations that are more sensitive to common features, a hallmark of efficient coding. This effect occurs in systems with otherwise unconstrained coding resources, and additionally when learning towards both supervised and unsupervised objectives. This result demonstrates that efficient codes can naturally emerge from gradient-like learning.

A web-based automated machine learning platform to analyze liquid biopsy data.

Liquid biopsy (LB) technologies continue to improve in sensitivity, specificity, and multiplexing and can measure an ever growing library of disease biomarkers. However, clinical interpretation of the increasingly large sets of data these technologies generate remains a challenge. Machine learning is a popular approach to discover and detect signatures of disease. However, limited machine learning expertise in the LB field has kept the discipline from fully leveraging these tools and risks improper analyses and irreproducible results. In this paper, we develop a web-based automated machine learning tool tailored specifically for LB, where machine learning models can be built without the user's input. We also incorporate a differential privacy algorithm, designed to limit the effects of overfitting that can arise from users iteratively developing a panel with feedback from our platform. We validate our approach by performing a meta-analysis on 11 published LB datasets, and found that we had similar or better performance compared to those reported in the literature. Moreover, we show that our platform's performance improved when incorporating information from prior LB datasets, suggesting that this approach can continue to improve with increased access to LB data. Finally, we show that by using our platform the results achieved in the literature can be matched using 40% of the number of subjects in the training set, potentially reducing study cost and time. This self-improving and overfitting-resistant automatic machine learning platform provides a new standard that can be used to validate machine learning works in the LB field.

Machine Learning for Neural Decoding.

Despite rapid advances in machine learning tools, the majority of neural decoding approaches still use traditional methods. Modern machine learning tools, which are versatile and easy to use, have the potential to significantly improve decoding performance. This tutorial describes how to effectively apply these algorithms for typical decoding problems. We provide descriptions, best practices, and code for applying common machine learning methods, including neural networks and gradient boosting. We also provide detailed comparisons of the performance of various methods at the task of decoding spiking activity in motor cortex, somatosensory cortex, and hippocampus. Modern methods, particularly neural networks and ensembles, significantly outperform traditional approaches, such as Wiener and Kalman filters. Improving the performance of neural decoding algorithms allows neuroscientists to better understand the information contained in a neural population and can help to advance engineering applications such as brain-machine interfaces. Our code package is available at github.com/kordinglab/neural_decoding.

The roles of supervised machine learning in systems neuroscience.

Over the last several years, the use of machine learning (ML) in neuroscience has been rapidly increasing. Here, we review ML's contributions, both realized and potential, across several areas of systems neuroscience. We describe four primary roles of ML within neuroscience: (1) creating solutions to engineering problems, (2) identifying predictive variables, (3) setting benchmarks for simple models of the brain, and (4) serving itself as a model for the brain. The breadth and ease of its applicability suggests that machine learning should be in the toolbox of most systems neuroscientists.

Modern Machine Learning as a Benchmark for Fitting Neural Responses.

Neuroscience has long focused on finding encoding models that effectively ask "what predicts neural spiking?" and generalized linear models (GLMs) are a typical approach. It is often unknown how much of explainable neural activity is captured, or missed, when fitting a model. Here we compared the predictive performance of simple models to three leading machine learning methods: feedforward neural networks, gradient boosted trees (using XGBoost), and stacked ensembles that combine the predictions of several methods. We predicted spike counts in macaque motor (M1) and somatosensory (S1) cortices from standard representations of reaching kinematics, and in rat hippocampal cells from open field location and orientation. Of these methods, XGBoost and the ensemble consistently produced more accurate spike rate predictions and were less sensitive to the preprocessing of features. These methods can thus be applied quickly to detect if feature sets relate to neural activity in a manner not captured by simpler methods. Encoding models built with a machine learning approach accurately predict spike rates and can offer meaningful benchmarks for simpler models.

Polymer Conjugation as a Strategy for Long-Range Order in Supramolecular Polymers.

Supramolecular polymers are polymers in which the individual subunits self-assemble via noncovalent and reversible bonds. An important axis of control for systems of mixed subunit composition is the order in which the subunit types assemble. Existing ordering techniques, which rely on pairwise interactions through the inclusion of highly specific chemistry, have the downside that patterns of length n require n specific chemistries, making long-range order complicated to attain. Here we present a simple alternative method: we attach varying numbers of polymers to self-assembling subunits, in our case ring shaped macrocycles, and the polymers' aversion to confinement imposes system order. We evaluate the feasibility of the strategy using coarse-grained molecular dynamics simulations of polymer-conjugated rings designed to model cyclic peptide nanotubes. We discuss the effects of polymer conjugation on the energetics of association and predict the equilibrium orderings for various ratios of ring types. The emergent patterns are associated with a certain stochastic disorder, which we quantify by deriving and employing a formula for the expected statistical weight of any pattern within the ensemble of all possible orderings.

Regulating Ion Transport in Peptide Nanotubes by Tailoring the Nanotube Lumen Chemistry.

We use atomistic nonequilibrium molecular dynamics simulations to demonstrate how specific ionic flux in peptide nanotubes can be regulated by tailoring the lumen chemistry through single amino acid substitutions. By varying the size and polarity of the functional group inserted into the nanotube interior, we are able to adjust the Na(+) flux by over an order of magnitude. Cl(-) is consistently denied passage. Bulky, nonpolar groups encourage interactions between the Na(+) and the peptide backbone carbonyl groups, disrupting the Na(+) solvation shell and slowing the transport of Na(+). Small groups have the opposite effect and accelerate flow. These results suggest that relative ion flux and selectivity can be precisely regulated in subnanometer pores by molecularly defining the lumen according to biological principles.