optimization

Modulating the Action of Cylindrical Proteases to Eliminate Neisseria Gonorrhea and Chlamydia Trachomatis Infections

Project Summary/Abstract Sexually transmitted bacteria diseases caused by Chlamydia trachomatis (Ctr) and Neisseria gonorrhoeae (NG) are the two most common sexually transmitted bacterial diseases. The infections caused by these pathogens may result in infertility, ectopic pregnancy, blindness, and perinatal mortality. Over 1.70 M cases of chlamydia and 0.65 M cases of drug-resistant gonorrhea are reported yearly in the US. Women with gonorrhea are co- infected with chlamydia in 17.6%–57.9% of cases, while women with chlamydia are co-infected with gonorrhea in 2.1%–17.2% of cases. These infections are treated with broad spectrum antibiotics, which can favor the development of resistance on NG/CTr but also in other bacteria, or damage the microbiota, diminishing its protective function and allowing bacteria and viruses to infect the patient. The Caseinolytic protease (ClpP) proteolytic machinery regulates protein turnover and homeostasis and is key in bacterial growth and development The machinery consists of the proteolytic unit (the ClpP) and its chaperone (ClpX), which transports proteins to be degraded, and it is termed the ClpXP. Our theory is that molecules that inhibit the action of the ClpX chaperone can become efficient antibacterial agents against both pathogens. We have found that the dihydrothiazepines can erradicate both pathogens and prevent the action of the ClpXP complex. Our goal is to advance the dihydrothiazepines as selective agents against Ctr and NG infections. To develop these therapeutic agents, we have envisioned four specific aims. Specific Aim 1. Synthesis and Optimization of the Pharmacophore. Our goal is to use computational models to design dihydrothiazepines molecule that will be synthesized, purified, and characterized using chemical techniques. The molecules will be tested against Ctr and NG and their toxicity against human cells evaluated. Also, we will determine their effect in other bacterial, including those from the microbiota. Specific Aim 2. Assessment of Stability and In Vivo Activity. We will study the stability of the most active molecules under various conditions. Then, we will study the pharmacokinetics, biodistribution , and antibacterial activity against Ctr and NG in mice. Specific Aim 3. Target Validation and Effect. We will study the ability of the compounds to inhibit the activity of ClpX using a luciferase assay and to block protein degradation. We will try grow crystal of the protein and the molecule and will study if the molecules prevent the assembly of the ClpXP system. Finally, we will assess the ability of the bacteria to develop resistance to the molecules.

Research on End-user Acceptability.and Long-term Impacts of HIV Cure Strategies (REALISE)

ABSTRACT Despite remarkable advances in HIV cure science, emerging cure candidates will likely involve trade-offs (e.g., incomplete eradication, monitoring burdens) and must compete with increasingly convenient long-acting ART; without early implementation guidance, even efficacious products may see limited uptake, particularly among the ~30–40% of people with HIV (PWH) in the U.S. who are not durably suppressed. We propose REALISE, a multidisciplinary program to define plausible cure profiles, quantify end-user preferences, and project population-level impact to inform product design and policy before market entry. Aim 1 conducts qualitative interviews with ~30 researchers and developers to delineate credible 10–20-year cure and long-acting treatment scenarios (eradication vs functional control, safety, monitoring, durability), yielding bounded “target product profiles.” Aim 2 elicits patient-centered preferences through a two-stage study: formative interviews (n=60; ≥50% not virally suppressed) to identify salient attributes; best-worst scaling (n=360 across Missouri, Georgia, and San Francisco) to prioritize attributes; and a discrete choice experiment (n=360) to quantify trade-offs versus alternative therapies, with latent class analysis to identify preference segments and estimate potential reach. Aim 3 integrates preference-based uptake from Aim 2 with Aim 1 efficacy and cost inputs in a mathematical model to estimate health impact, QALYs, net QALYs, and incremental cost-effectiveness across heterogeneous populations and Ending the HIV Epidemic jurisdictions. Innovation lies in linking cure R&D horizons to end-user preferences and transmission-dynamic outcomes, an approach that anticipates real-world use rather than retrofitting after approval. Deliverables include ranked cure attributes for product optimization, uptake projections including among unsuppressed PWH, and jurisdiction-specific value assessments to guide public health investment. By aligning cure design with what patients will accept and systems can sustain, REALISE will accelerate effective deployment of future cure strategies and maximize their contribution to Ending the HIV Epidemic. In doing so, this study advances NIH's priorities by connecting implementation science with prevention, treatment, and cure research. Using a multidisciplinary strategy to refine and extend `target product profiles,' REALISE will ensure cure development reflects patient needs and accelerate translation into real-world benefit.

Multimodal computational models for early prediction of peritoneal recurrence in gastric cancer

ABSTRACT Gastric cancer represents a significant disease burden and is a leading cause of cancer-related deaths in the United States and globally. Approximately 80% of gastric cancer patients are diagnosed at an advanced stage, with the peritoneum being the most common site of relapse (peritoneal recurrence) after radical surgery. Nearly 50% of patients with advanced-stage gastric cancer develop peritoneal recurrence post-surgery, resulting in a median survival of only 3–6 months and a markedly reduced quality of life. Early peritoneal recurrence is primarily characterized by micro-metastasis, which traditional imaging techniques struggle to detect due to the small size of metastatic nodules. Predicting the likelihood and timing of peritoneal recurrence is crucial for identifying at- risk patients, enabling timely interventions that could improve survival rates and quality of life. Unfortunately, reliable predictive biomarkers and models for peritoneal recurrence in gastric cancer are lacking in clinical practice, highlighting an urgent need for innovative predictive tools. This proposal aims to develop and validate novel predictive models for early peritoneal recurrence in gastric cancer, leveraging advanced deep learning techniques and multimodal integration of clinical, radiological (CT), and histopathological (hematoxylin and eosin, H&E) data. In Aim 1, we will develop a rational approach for predicting peritoneal recurrence by creating a novel deep learning multimodal method guided by genomics knowledge. Additionally, we will integrate both deep learning-extracted features and traditional hand-crafted radiomics features with clinical data to improve prediction accuracy. Aim 2 focuses on developing a robust prediction model of peritoneal recurrence utilizing a pre-trained foundation model from large-scale H&E image data. Aim 3 will combine CT, H&E, and clinical data to further enhance predictive capabilities, employing an innovative cross-modal collaborative optimization approach for multimodal data integration. All models will be trained and internally validated using a retrospective cohort from Atrium Health Wake Forest Baptist Comprehensive Cancer Center and externally validated in two independent cohorts from additional institutions to ensure robustness across populations and imaging protocols. Additionally, we will compare our models with existing methods, including clinical staging and alternative fusion strategies. If successful, these models will enhance risk stratification and prediction of peritoneal recurrence in gastric cancer patients, significantly improving survival rates and quality of life by identifying those likely to develop peritoneal recurrence post-surgery and facilitating timely intervention. Furthermore, they can help avoid the risk of complications and extra medical costs associated with overtreatment. Since the information is derived from routinely examined CT, H&E and clinical data, they could be seamlessly integrated into current clinical workflows. The AI technology developed through this project has the potential to benefit underserved populations in low- resource settings and reduce healthcare disparities in the U.S.

Optimization of a novel and effective antiviral agent targeting Zika NS4B

This project focuses on developing novel anti-Zika virus (ZIKV) compounds targeting the NS4B protein, which is crucial for viral replication. ZIKV poses a significant medical challenge due to its potential for severe pathogenic outcomes, such as congenital Zika syndrome and Guillain-Barré Syndrome. Furthermore, its pandemic potential has been increasing with the expansion of carrier mosquito habitats. The project aims to address the urgent need for anti-ZIKV therapeutics that could greatly reduce severity of symptoms and minimize vertical and community transmissions. We have identified a novel small-molecule series with a benzamide scaffold through a cell-based, antiviral ultra-high-throughput screen. This series demonstrates strong potency against ZIKV without measurable cytotoxicity or non-specific antiviral effects, justifying this scaffold as a lead series for further development. Preliminary mechanism-of-action studies, utilizing genetic, biochemical, and virological assays, suggest that this series may inhibit the formation of the ZIKV viral replicase complex by interfering with NS4B. Our goal for this project is to develop a preclinical therapeutic candidate for ZIKV that demonstrates promising therapeutic activity following oral administration in ZIKV-infected mice, at a dosage that shows no clinical toxicity. The project has the following significant and novel objectives: 1) Optimize the benzamide lead for potency and drug-likeness; 2) Develop a lead candidate and a backup compound with optimized pharmacokinetic, pharmacodynamic, and toxicity profiles; 3) Determine the molecular mechanisms of action of the benzamide series using novel structural approaches to assist medicinal chemistry studies; 4) Evaluate the in vivo therapeutic efficacy and safety in mouse models and develop the best therapeutic regime. This project seeks to develop effective antivirals for ZIKV with high retention in the blood and central nervous system (CNS) and high oral bioavailability. The expected successful outcomes will provide significant advancements in ZIKV therapeutics and open new avenues for treating other flavivirus infections

Implementing a New Paradigm for Antifungal Drug Development

About 30% of the drugs currently in clinical use function through covalent modification of their target. Yet, until recently, none of these covalent drugs were specifically designed to utilize this irreversible mode of action. It is our hypothesis that the production of a new class of covalent inactivators, designed to selectively modify new drug targets, will lead to novel agents with efficacy against both native and drug-resistant pathogenic fungal species. Because of their novelty these agents will also offer a greater opportunity to bypass the existing mechanisms of drug resistance. Pathogenic fungal infections remain among the leading causes of human mortality, and this threat is rising due to the increasing prevalence of drug- resistance strains and the paucity of effective antifungal drugs against the more virulent fungal species. Our proposed new drug target is an enzyme that plays a critical role in a uniquely microbial pathway that is essential for the survival of fungal organisms. To test our hypothesis and achieve the goals of this project we plan to complete the following specific aims during the initial R21 phase of this project: (1) Optimization of the potency of novel enzyme inactivators. Our goals here are to use our strong preliminary results to address critical barriers that must be overcome to convert potent enzyme inactivators into advanced drug candidates, thereby achieving higher target selectivity and increasing compound reactivity once bound to the target; (2) Enhance the antifungal capability of these enzyme inactivators. Our strategy for this aim is focused on the incorporation of conjugate partners into this new class of covalent inactivators, enabling them to potentially utilize the existing nutrient uptake systems to achieve toxic levels in Candida species; (3) Examine the target selectivity of our new antifungal agents. Results from fungal growth inhibition and fungal killing assays will be used to evaluate and rank the efficacy of our compounds against both wild-type and drug-resistant Candida strains. Specific milestones are presented to evaluate our achievement of these initial aims. Once accomplished we will immediately proceed to the R33 phase of this project, with the aims of: (4) Pharmacological evaluation of lead candidates, though ranking the drug candidates based on their ADME, pharmacokinetic and toxicity properties; and then (5) Evaluate the efficacy of our candidates against pathogenic fungal infections. A systematic infection animal model will be utilized for candidate screening to identify the best agents against disseminated fungal infections, followed by further efficacy screening in an oral infection model. Completion of these aims will produce, refine and evaluate a new class of antifungal agents with a novel mode of action against an unexplored but essential fungal target. The agents with the most promising drug profiles will then be moved into advanced preclinical trials used to select the most effective new antifungal agents.

A PROTAC Strategy to Combat Botulinum Neurotoxicity

PROJECT SUMMARY/ABSTRACT Botulinum neurotoxin (BoNT), the causative agent of botulism, is the most potent toxin known to humans. While BoNTs are widely recognized for their therapeutic and cosmetic applications, such as Botox™, their increasing use has raised concerns about iatrogenic botulism. Due to their extreme lethality, ease of production, and history of weaponization, the Centers for Disease Control and Prevention (CDC) classifies BoNTs as a Category A bioterrorism threat. Among the seven major serotypes (A-G), BoNT/A, BoNT/B, and BoNT/E account for over 95% of human botulism cases with A being the most prevalent. Despite the severity of botulism, no approved therapeutic exists to rescue intoxicated neurons. The current treatment, a heptavalent antitoxin, can only slow disease progression and requires early administration and prolonged hospitalization due to the inability of antibodies to penetrate infected cells. In the field of small- molecule inhibitors (SMIs), promising scaffolds targeting BoNT/A have been discovered, offering opportunities for further derivatization to incorporate bifunctional approaches. Developing a clinically viable therapeutic requires inhibiting the zinc (Zn2+) metalloprotease light chain (LC) as well as addressing toxin persistence. Through extensive inhibitor screening, we have identified two classes of small molecules that inhibit BoNT/A with submicromolar affinity and demonstrate efficacy in both cellular and animal models. However, the transient nature of these inhibitors necessitates the need of a sustained clearance approach. To achieve this, we propose integrating our previously identified BoNT/A LC SMIs with a targeted protein degradation (TPD) technology for toxin elimination. Based upon the background outlined, vide supra, our research strategy for the ablation of BoNT/A will be focused upon the following three specific objectives: 1) Structural Optimization – Utilize molecular docking, and structure-activity relationship (SAR) analysis to modify inhibitors for TPD ligand attachment. 2) Degrader Design – Development of ubiquitin-protease system (UPS)-based proteolysis-targeting chimeras (PROTACs) and autophagy-targeting chimeras to enhance degradation efficiency. 3) Cellular Evaluation – Assess enzyme inhibition, toxin clearance, degradation kinetics in cells.

Neural circuits underlying sleep structure and functions

Sleep is an active state critical for processing emotional memories encoded during waking in both humans and animals. There is a remarkable overlap between the brain structures and circuits active during sleep, particularly rapid eye-movement (REM) sleep, and the those encoding emotions. Accordingly, disruptions in sleep quality or quantity, including REM sleep, are often associated with, and precede the onset of, nearly all affective psychiatric and mood disorders. In this context, a major biomedical challenge is to better understand the underlying mechanisms of the relationship between (REM) sleep and emotion encoding to improve treatments for mental health. This lecture will summarize our investigation of the cellular and circuit mechanisms underlying sleep architecture, sleep oscillations, and local brain dynamics across sleep-wake states using electrophysiological recordings combined with single-cell calcium imaging or optogenetics. The presentation will detail the discovery of a 'somato-dendritic decoupling'in prefrontal cortex pyramidal neurons underlying REM sleep-dependent stabilization of optimal emotional memory traces. This decoupling reflects a tonic inhibition at the somas of pyramidal cells, occurring simultaneously with a selective disinhibition of their dendritic arbors selectively during REM sleep. Recent findings on REM sleep-dependent subcortical inputs and neuromodulation of this decoupling will be discussed in the context of synaptic plasticity and the optimization of emotional responses in the maintenance of mental health.

Brain circuits for spatial navigation

In this webinar on spatial navigation circuits, three researchers—Ann Hermundstad, Ila Fiete, and Barbara Webb—discussed how diverse species solve navigation problems using specialized yet evolutionarily conserved brain structures. Hermundstad illustrated the fruit fly’s central complex, focusing on how hardwired circuit motifs (e.g., sinusoidal steering curves) enable rapid, flexible learning of goal-directed navigation. This framework combines internal heading representations with modifiable goal signals, leveraging activity-dependent plasticity to adapt to new environments. Fiete explored the mammalian head-direction system, demonstrating how population recordings reveal a one-dimensional ring attractor underlying continuous integration of angular velocity. She showed that key theoretical predictions—low-dimensional manifold structure, isometry, uniform stability—are experimentally validated, underscoring parallels to insect circuits. Finally, Webb described honeybee navigation, featuring path integration, vector memories, route optimization, and the famous waggle dance. She proposed that allocentric velocity signals and vector manipulation within the central complex can encode and transmit distances and directions, enabling both sophisticated foraging and inter-bee communication via dance-based cues.

Applied cognitive neuroscience to improve learning and therapeutics

Advancements in cognitive neuroscience have provided profound insights into the workings of the human brain and the methods used offer opportunities to enhance performance, cognition, and mental health. Drawing upon interdisciplinary collaborations in the University of California San Diego, Human Performance Optimization Lab, this talk explores the application of cognitive neuroscience principles in three domains to improve human performance and alleviate mental health challenges. The first section will discuss studies addressing the role of vision and oculomotor function in athletic performance and the potential to train these foundational abilities to improve performance and sports outcomes. The second domain considers the use of electrophysiological measurements of the brain and heart to detect, and possibly predict, errors in manual performance, as shown in a series of studies with surgeons as they perform robot-assisted surgery. Lastly, findings from clinical trials testing personalized interventional treatments for mood disorders will be discussed in which the temporal and spatial parameters of transcranial magnetic stimulation (TMS) are individualized to test if personalization improves treatment response and can be used as predictive biomarkers to guide treatment selection. Together, these translational studies use the measurement tools and constructs of cognitive neuroscience to improve human performance and well-being.

Maintaining Plasticity in Neural Networks

Nonstationarity presents a variety of challenges for machine learning systems. One surprising pathology which can arise in nonstationary learning problems is plasticity loss, whereby making progress on new learning objectives becomes more difficult as training progresses. Networks which are unable to adapt in response to changes in their environment experience plateaus or even declines in performance in highly non-stationary domains such as reinforcement learning, where the learner must quickly adapt to new information even after hundreds of millions of optimization steps. The loss of plasticity manifests in a cluster of related empirical phenomena which have been identified by a number of recent works, including the primacy bias, implicit under-parameterization, rank collapse, and capacity loss. While this phenomenon is widely observed, it is still not fully understood. This talk will present exciting recent results which shed light on the mechanisms driving the loss of plasticity in a variety of learning problems and survey methods to maintain network plasticity in non-stationary tasks, with a particular focus on deep reinforcement learning.

Current and future trends in neuroimaging

With the advent of several different fMRI analysis tools and packages outside of the established ones (i.e., SPM, AFNI, and FSL), today's researcher may wonder what the best practices are for fMRI analysis. This talk will discuss some of the recent trends in neuroimaging, including design optimization and power analysis, standardized analysis pipelines such as fMRIPrep, and an overview of current recommendations for how to present neuroimaging results. Along the way we will discuss the balance between Type I and Type II errors with different correction mechanisms (e.g., Threshold-Free Cluster Enhancement and Equitable Thresholding and Clustering), as well as considerations for working with large open-access databases.

Signatures of criticality in efficient coding networks

The critical brain hypothesis states that the brain can benefit from operating close to a second-order phase transition. While it has been shown that several computational aspects of sensory information processing (e.g., sensitivity to input) are optimal in this regime, it is still unclear whether these computational benefits of criticality can be leveraged by neural systems performing behaviorally relevant computations. To address this question, we investigate signatures of criticality in networks optimized to perform efficient encoding. We consider a network of leaky integrate-and-fire neurons with synaptic transmission delays and input noise. Previously, it was shown that the performance of such networks varies non-monotonically with the noise amplitude. Interestingly, we find that in the vicinity of the optimal noise level for efficient coding, the network dynamics exhibits signatures of criticality, namely, the distribution of avalanche sizes follows a power law. When the noise amplitude is too low or too high for efficient coding, the network appears either super-critical or sub-critical, respectively. This result suggests that two influential, and previously disparate theories of neural processing optimization—efficient coding, and criticality—may be intimately related

Understanding Machine Learning via Exactly Solvable Statistical Physics Models

The affinity between statistical physics and machine learning has a long history. I will describe the main lines of this long-lasting friendship in the context of current theoretical challenges and open questions about deep learning. Theoretical physics often proceeds in terms of solvable synthetic models, I will describe the related line of work on solvable models of simple feed-forward neural networks. I will highlight a path forward to capture the subtle interplay between the structure of the data, the architecture of the network, and the optimization algorithms commonly used for learning.

Spatially-embedded recurrent neural networks reveal widespread links between structural and functional neuroscience findings

Brain networks exist within the confines of resource limitations. As a result, a brain network must overcome metabolic costs of growing and sustaining the network within its physical space, while simultaneously implementing its required information processing. To observe the effect of these processes, we introduce the spatially-embedded recurrent neural network (seRNN). seRNNs learn basic task-related inferences while existing within a 3D Euclidean space, where the communication of constituent neurons is constrained by a sparse connectome. We find that seRNNs, similar to primate cerebral cortices, naturally converge on solving inferences using modular small-world networks, in which functionally similar units spatially configure themselves to utilize an energetically-efficient mixed-selective code. As all these features emerge in unison, seRNNs reveal how many common structural and functional brain motifs are strongly intertwined and can be attributed to basic biological optimization processes. seRNNs can serve as model systems to bridge between structural and functional research communities to move neuroscientific understanding forward.

Universal function approximation in balanced spiking networks through convex-concave boundary composition

The spike-threshold nonlinearity is a fundamental, yet enigmatic, component of biological computation — despite its role in many theories, it has evaded definitive characterisation. Indeed, much classic work has attempted to limit the focus on spiking by smoothing over the spike threshold or by approximating spiking dynamics with firing-rate dynamics. Here, we take a novel perspective that captures the full potential of spike-based computation. Based on previous studies of the geometry of efficient spike-coding networks, we consider a population of neurons with low-rank connectivity, allowing us to cast each neuron’s threshold as a boundary in a space of population modes, or latent variables. Each neuron divides this latent space into subthreshold and suprathreshold areas. We then demonstrate how a network of inhibitory (I) neurons forms a convex, attracting boundary in the latent coding space, and a network of excitatory (E) neurons forms a concave, repellant boundary. Finally, we show how the combination of the two yields stable dynamics at the crossing of the E and I boundaries, and can be mapped onto a constrained optimization problem. The resultant EI networks are balanced, inhibition-stabilized, and exhibit asynchronous irregular activity, thereby closely resembling cortical networks of the brain. Moreover, we demonstrate how such networks can be tuned to either suppress or amplify noise, and how the composition of inhibitory convex and excitatory concave boundaries can result in universal function approximation. Our work puts forth a new theory of biologically-plausible computation in balanced spiking networks, and could serve as a novel framework for scalable and interpretable computation with spikes.

Potential pathways for novel interventions in TLE

Inhibition of seizures can come from expected – and surprising – sources. In this talk I will explore circuit elements, both within and external to the temporal lobe, which may be able inhibit hippocampal seizures, and how specific aspects of intervention strategies can be critical for outcomes. We’ll discuss novel sources of inhibition within the hippocampus, the cerebellum as a potential target, and closed-loop optimization of stimulation parameters

What does time of day mean for vision?

Profound changes in the visual environment occur over the course of the day-night cycle. There is therefore a profound pressure for cells and circuits within the visual system to adjust their function over time, to match the prevailing visual environment. Here, I will discuss electrophysiological data collected from nocturnal and diurnal rodents that reveal how the visual code is ‘temporally optimised’ by 1) the retina’s circadian clock, and 2) a change in behavioural temporal niche.

Optimization at the Single Neuron Level:​ Prediction of Spike Sequences and Emergence of Synaptic Plasticity Mechanisms

Intelligent behavior depends on the brain’s ability to anticipate future events. However, the learning rules that enable neurons to predict and fire ahead of sensory inputs remain largely unknown. We propose a plasticity rule based on pre-dictive processing, where the neuron learns a low-rank model of the synaptic input dynamics in its membrane potential. Neurons thereby amplify those synapses that maximally predict other synaptic inputs based on their temporal relations, which provide a solution to an optimization problem that can be implemented at the single-neuron level using only local information. Consequently, neurons learn sequences over long timescales and shift their spikes towards the first inputs in a sequence. We show that this mechanism can explain the development of anticipatory motion signaling and recall in the visual system. Furthermore, we demonstrate that the learning rule gives rise to several experimentally observed STDP (spike-timing-dependent plasticity) mechanisms. These findings suggest prediction as a guiding principle to orchestrate learning and synaptic plasticity in single neurons.

NMC4 Short Talk: Hypothesis-neutral response-optimized models of higher-order visual cortex reveal strong semantic selectivity

Modeling neural responses to naturalistic stimuli has been instrumental in advancing our understanding of the visual system. Dominant computational modeling efforts in this direction have been deeply rooted in preconceived hypotheses. In contrast, hypothesis-neutral computational methodologies with minimal apriorism which bring neuroscience data directly to bear on the model development process are likely to be much more flexible and effective in modeling and understanding tuning properties throughout the visual system. In this study, we develop a hypothesis-neutral approach and characterize response selectivity in the human visual cortex exhaustively and systematically via response-optimized deep neural network models. First, we leverage the unprecedented scale and quality of the recently released Natural Scenes Dataset to constrain parametrized neural models of higher-order visual systems and achieve novel predictive precision, in some cases, significantly outperforming the predictive success of state-of-the-art task-optimized models. Next, we ask what kinds of functional properties emerge spontaneously in these response-optimized models? We examine trained networks through structural ( feature visualizations) as well as functional analysis (feature verbalizations) by running `virtual' fMRI experiments on large-scale probe datasets. Strikingly, despite no category-level supervision, since the models are solely optimized for brain response prediction from scratch, the units in the networks after optimization act as detectors for semantic concepts like `faces' or `words', thereby providing one of the strongest evidences for categorical selectivity in these visual areas. The observed selectivity in model neurons raises another question: are the category-selective units simply functioning as detectors for their preferred category or are they a by-product of a non-category-specific visual processing mechanism? To investigate this, we create selective deprivations in the visual diet of these response-optimized networks and study semantic selectivity in the resulting `deprived' networks, thereby also shedding light on the role of specific visual experiences in shaping neuronal tuning. Together with this new class of data-driven models and novel model interpretability techniques, our study illustrates that DNN models of visual cortex need not be conceived as obscure models with limited explanatory power, rather as powerful, unifying tools for probing the nature of representations and computations in the brain.

Deep kernel methods

Deep neural networks (DNNs) with the flexibility to learn good top-layer representations have eclipsed shallow kernel methods without that flexibility. Here, we take inspiration from deep neural networks to develop a new family of deep kernel method. In a deep kernel method, there is a kernel at every layer, and the kernels are jointly optimized to improve performance (with strong regularisation). We establish the representational power of deep kernel methods, by showing that they perform exact inference in an infinitely wide Bayesian neural network or deep Gaussian process. Next, we conjecture that the deep kernel machine objective is unimodal, and give a proof of unimodality for linear kernels. Finally, we exploit the simplicity of the deep kernel machine loss to develop a new family of optimizers, based on a matrix equation from control theory, that converges in around 10 steps.

Norse: A library for gradient-based learning in Spiking Neural Networks

We introduce Norse: An open-source library for gradient-based training of spiking neural networks. In contrast to neuron simulators which mainly target computational neuroscientists, our library seamlessly integrates with the existing PyTorch ecosystem using abstractions familiar to the machine learning community. This has immediate benefits in that it provides a familiar interface, hardware accelerator support and, most importantly, the ability to use gradient-based optimization. While many parallel efforts in this direction exist, Norse emphasizes flexibility and usability in three ways. Users can conveniently specify feed-forward (convolutional) architectures, as well as arbitrarily connected recurrent networks. We strictly adhere to a functional and class-based API such that neuron primitives and, for example, plasticity rules composes. Finally, the functional core API ensures compatibility with the PyTorch JIT and ONNX infrastructure. We have made progress to support network execution on the SpiNNaker platform and plan to support other neuromorphic architectures in the future. While the library is useful in its present state, it also has limitations we will address in ongoing work. In particular, we aim to implement event-based gradient computation, using the EventProp algorithm, which will allow us to support sparse event-based data efficiently, as well as work towards support of more complex neuron models. With this library, we hope to contribute to a joint future of computational neuroscience and neuromorphic computing.

Event-based Backpropagation for Exact Gradients in Spiking Neural Networks

Gradient-based optimization powered by the backpropagation algorithm proved to be the pivotal method in the training of non-spiking artificial neural networks. At the same time, spiking neural networks hold the promise for efficient processing of real-world sensory data by communicating using discrete events in continuous time. We derive the backpropagation algorithm for a recurrent network of spiking (leaky integrate-and-fire) neurons with hard thresholds and show that the backward dynamics amount to an event-based backpropagation of errors through time. Our derivation uses the jump conditions for partial derivatives at state discontinuities found by applying the implicit function theorem, allowing us to avoid approximations or substitutions. We find that the gradient exists and is finite almost everywhere in weight space, up to the null set where a membrane potential is precisely tangent to the threshold. Our presented algorithm, EventProp, computes the exact gradient with respect to a general loss function based on spike times and membrane potentials. Crucially, the algorithm allows for an event-based communication scheme in the backward phase, retaining the potential advantages of temporal sparsity afforded by spiking neural networks. We demonstrate the optimization of spiking networks using gradients computed via EventProp and the Yin-Yang and MNIST datasets with either a spike time-based or voltage-based loss function and report competitive performance. Our work supports the rigorous study of gradient-based optimization in spiking neural networks as well as the development of event-based neuromorphic architectures for the efficient training of spiking neural networks. While we consider the leaky integrate-and-fire model in this work, our methodology generalises to any neuron model defined as a hybrid dynamical system.

Optimal initialization strategies for Deep Spiking Neural Networks

Recent advances in neuromorphic hardware and Surrogate Gradient (SG) learning highlight the potential of Spiking Neural Networks (SNNs) for energy-efficient signal processing and learning. Like in Artificial Neural Networks (ANNs), training performance in SNNs strongly depends on the initialization of synaptic and neuronal parameters. While there are established methods of initializing deep ANNs for high performance, effective strategies for optimal SNN initialization are lacking. Here, we address this gap and propose flexible data-dependent initialization strategies for SNNs.

Optimising spiking interneuron circuits for compartment-specific feedback

Cortical circuits process information by rich recurrent interactions between excitatory neurons and inhibitory interneurons. One of the prime functions of interneurons is to stabilize the circuit by feedback inhibition, but the level of specificity on which inhibitory feedback operates is not fully resolved. We hypothesized that inhibitory circuits could enable separate feedback control loops for different synaptic input streams, by means of specific feedback inhibition to different neuronal compartments. To investigate this hypothesis, we adopted an optimization approach. Leveraging recent advances in training spiking network models, we optimized the connectivity and short-term plasticity of interneuron circuits for compartment-specific feedback inhibition onto pyramidal neurons. Over the course of the optimization, the interneurons diversified into two classes that resembled parvalbumin (PV) and somatostatin (SST) expressing interneurons. The resulting circuit can be understood as a neural decoder that inverts the nonlinear biophysical computations performed within the pyramidal cells. Our model provides a proof of concept for studying structure-function relations in cortical circuits by a combination of gradient-based optimization and biologically plausible phenomenological models

On the implicit bias of SGD in deep learning

Tali's work emphasized the tradeoff between compression and information preservation. In this talk I will explore this theme in the context of deep learning. Artificial neural networks have recently revolutionized the field of machine learning. However, we still do not have sufficient theoretical understanding of how such models can be successfully learned. Two specific questions in this context are: how can neural nets be learned despite the non-convexity of the learning problem, and how can they generalize well despite often having more parameters than training data. I will describe our recent work showing that gradient-descent optimization indeed leads to 'simpler' models, where simplicity is captured by lower weight norm and in some cases clustering of weight vectors. We demonstrate this for several teacher and student architectures, including learning linear teachers with ReLU networks, learning boolean functions and learning convolutional pattern detection architectures.

Credit Assignment in Neural Networks through Deep Feedback Control

The success of deep learning sparked interest in whether the brain learns by using similar techniques for assigning credit to each synaptic weight for its contribution to the network output. However, the majority of current attempts at biologically-plausible learning methods are either non-local in time, require highly specific connectivity motives, or have no clear link to any known mathematical optimization method. Here, we introduce Deep Feedback Control (DFC), a new learning method that uses a feedback controller to drive a deep neural network to match a desired output target and whose control signal can be used for credit assignment. The resulting learning rule is fully local in space and time and approximates Gauss-Newton optimization for a wide range of feedback connectivity patterns. To further underline its biological plausibility, we relate DFC to a multi-compartment model of cortical pyramidal neurons with a local voltage-dependent synaptic plasticity rule, consistent with recent theories of dendritic processing. By combining dynamical system theory with mathematical optimization theory, we provide a strong theoretical foundation for DFC that we corroborate with detailed results on toy experiments and standard computer-vision benchmarks.

Advances in Computational Psychiatry: Understanding (cognitive) control as a network process

The human brain is a complex organ characterized by heterogeneous patterns of interconnections. Non-invasive imaging techniques now allow for these patterns to be carefully and comprehensively mapped in individual humans, paving the way for a better understanding of how wiring supports cognitive processes. While a large body of work now focuses on descriptive statistics to characterize these wiring patterns, a critical open question lies in how the organization of these networks constrains the potential repertoire of brain dynamics. In this talk, I will describe an approach for understanding how perturbations to brain dynamics propagate through complex wiring patterns, driving the brain into new states of activity. Drawing on a range of disciplinary tools – from graph theory to network control theory and optimization – I will identify control points in brain networks and characterize trajectories of brain activity states following perturbation to those points. Finally, I will describe how these computational tools and approaches can be used to better understand the brain's intrinsic control mechanisms and their alterations in psychiatric conditions.

A computational explanation for domain specificity in the human brain

Many regions of the human brain conduct highly specific functions, such as recognizing faces, understanding language, and thinking about other people’s thoughts. Why might this domain specific organization be a good design strategy for brains, and what is the origin of domain specificity in the first place? In this talk, I will present recent work testing whether the segregation of face and object perception in human brains emerges naturally from an optimization for both tasks. We trained artificial neural networks on face and object recognition, and found that networks were able to perform both tasks well by spontaneously segregating them into distinct pathways. Critically, networks neither had prior knowledge nor any inductive bias about the tasks. Furthermore, networks optimized on tasks which apparently do not develop specialization in the human brain, such as food or cars, and object categorization showed less task segregation. These results suggest that functional segregation can spontaneously emerge without a task-specific bias, and that the domain-specific organization of the cortex may reflect a computational optimization for the real-world tasks humans solve.

E-prop: A biologically inspired paradigm for learning in recurrent networks of spiking neurons

Transformative advances in deep learning, such as deep reinforcement learning, usually rely on gradient-based learning methods such as backpropagation through time (BPTT) as a core learning algorithm. However, BPTT is not argued to be biologically plausible, since it requires to a propagate gradients backwards in time and across neurons. Here, we propose e-prop, a novel gradient-based learning method with local and online weight update rules for recurrent neural networks, and in particular recurrent spiking neural networks (RSNNs). As a result, e-prop has the potential to provide a substantial fraction of the power of deep learning to RSNNs. In this presentation, we will motivate e-prop from the perspective of recent insights in neuroscience and show how these have to be combined to form an algorithm for online gradient descent. The mathematical results will be supported by empirical evidence in supervised and reinforcement learning tasks. We will also discuss how limitations that are inherited from gradient-based learning methods, such as sample-efficiency, can be addressed by considering an evolution-like optimization that enhances learning on particular task families. The emerging learning architecture can be used to learn tasks by a single demonstration, hence enabling one-shot learning.

How brain evolutionary mechanisms could inspire AI structural designs

Across evolution and, in particular, in brain evolutionary development we can observe how diverse adaptive biological mechanisms are displayed as a solution to environmental demands. In this talk, I will discuss some examples of emerging evolutionary developmental strategies allowing to increase brain computational capacities and how neurodevelopmental conservation, divergence, and convergence would inspire AI systems optimization.

Spanning the arc between optimality theories and data

Ideas about optimization are at the core of how we approach biological complexity. Quantitative predictions about biological systems have been successfully derived from first principles in the context of efficient coding, metabolic and transport networks, evolution, reinforcement learning, and decision making, by postulating that a system has evolved to optimize some utility function under biophysical constraints. Yet as normative theories become increasingly high-dimensional and optimal solutions stop being unique, it gets progressively hard to judge whether theoretical predictions are consistent with, or "close to", data. I will illustrate these issues using efficient coding applied to simple neuronal models as well as to a complex and realistic biochemical reaction network. As a solution, we developed a statistical framework which smoothly interpolates between ab initio optimality predictions and Bayesian parameter inference from data, while also permitting statistically rigorous tests of optimality hypotheses.

Multiparameter optimization for whole-brain personalization

Bernstein Conference 2024

tDCS montage optimization for the treatment of epilepsy using Neurotwins

Bernstein Conference 2024

Modeling and optimization for neuromodulation in spinal cord stimulation

COSYNE 2022

Modeling and optimization for neuromodulation in spinal cord stimulation

COSYNE 2022

Online neural modeling and Bayesian optimization for closed-loop adaptive experiments

COSYNE 2022

Online neural modeling and Bayesian optimization for closed-loop adaptive experiments

COSYNE 2022

Optimization of error distributions as a design principle for neural representations

COSYNE 2022

Optimization of error distributions as a design principle for neural representations

COSYNE 2022

Top-down optimization recovers biological coding principles of single-neuron adaptation in RNNs

COSYNE 2022

Top-down optimization recovers biological coding principles of single-neuron adaptation in RNNs

COSYNE 2022

A data-driven biomechanical modeling and optimization pipeline for studying salamander locomotions

COSYNE 2025

Invariant synaptic density across species links functional stability and wiring optimization principles

COSYNE 2025

Safe Bayesian Optimization for High-Dimensional Neural Control of Movement

COSYNE 2025

Second-order forward-mode optimization of RNNs for neuroscience

COSYNE 2025

Conduction block stimulation optimization by envelope modulation toward the reduction of onset response

Optimization of Myelination at Mid-Age: Interaction Analysis between Grey Matter and White Matter compartments

Towards an optimization of functional localizers in non-human primate imaging with (fMRI) frequency-tagging

Advancing optogenetic hearing restoration through cross-modal optimization

FENS Forum 2024

Advanced metamodelling on the o2S2PARC computational neurosciences platform facilitates stimulation selectivity and power efficiency optimization and intelligent control

FENS Forum 2024

Chemogenetic inhibition of the lateral hypothalamic area to decrease food intake in rats: A proof-of-concept study and the optimization of technical parameters

FENS Forum 2024

Optimization of a multiplexed, cell-based assay of neuronal function

FENS Forum 2024

Optimization techniques for machine learning based classification involving large-scale neuroscience datasets

Neuromatch 5