Dimensionality reduction

…including high-dimensional geometry, dimensionality reduction, scalable algorithms, uncertainty and machine learning. The conference takes place in person at the Salt Palace Convention Center on 16–20 November 2026. Early rates run through 19 October; abstract and travel-support deadlines have passed. Registration also covers the co-located SIAM Imaging Science and Data Mining conferences.

…normalisation, batch correction, dimensionality reduction, clustering, annotation and differential expression. Later sessions cover trajectory and pseudotime inference, RNA velocity, single-cell multi-omics integration and functional analysis. Participants finish with an independent dataset analysis and presentation. Bash and R/Bioconductor support reproducible workflows and interpre

The dopaminergic system facilitates associative learning and motivation by signaling reward expectation and positive or negative valence in selected brain regions. Previous studies measuring somatic and axonal activity suggest a heterogeneous representation of reinforcement signaling by dopaminergic neurons, but it remains unclear how dopamine signals for expectation and valence interact in various brain regions. Here, we used an improved fluorescent dopamine sensor to record dopamine signals associated with reward predictions and valence in eight distinct striatal and non-striatal regions. We

Single neurons show complex responses during cognitive tasks and dimensionality reduction has been a successful approach for finding structure in heterogeneous neural activity. When applied to neural response patterns in a population state space, where each axis represents activity of one neuron, dimensionality reduction methods often reveal low-dimensional manifolds encoding behaviorally relevant variables. Alternatively, when the same responses are viewed in a selectivity space, where each axis represents a response feature (e.g., task condition or time), dimensionality reduction methods can

…level remains elusive due to the large dimensionality and nonlinearity of neural activity. Current dimensionality reduction t

…field has developed useful nonlinear dimensionality reduction methods such as t-distributed stochastic neighbor embedding (t-SNE) [1], their implementation in simple biological circuits remains unclear. Here, we develop a biologically plausible dimensionality reduction algorithm compatible with t-SNE. The proposed algorithm, named Hebbian

Recent studies have proposed that neural circuits have a $\textit{task manifold}$: i.e., a subset of the neural state-space to which neural activity is confined as an animal performs a task [1]. Thus, discovering and characterising these manifolds and their associated dynamics from experimental data can shed light on the neural computations unfolding within the brain during various cognitive tasks. Yet, common manifold discovery methods often do not take into account that neural data is generated by an underlying dynamical system. To address this, we first derive a general class of manifolds t

Most mammalian species present topological maps of preference for the ori- entation of edges in their early visual cortex. However, despite the evidence that evolution has favoured topological map organisation in most species, a clear explanation for the functional role of such maps is lacking. Traditional approaches based on wiring minimisation are incomplete, showing how topology can reduce metabolic costs, but only after assuming a functional role of short- ranged like-tuned connectivity. In this work, we claim that topological maps are for reducing the complexity of the propagated informat

In this study we explore network connectivity of in vitro neuronal systems of live biological cells within a real-time game environment of the game Pong. We use DishBrain - a system that embodies in vitro neural networks with in silico computation using a high-density multi-electrode array (MEA). We compare functional connectivity across resting-state sessions and gameplay sessions, during which the cells control a paddle able to hit a ball. The location of the paddle and ball are temporally and spatially encoded and presented to the neurons via electrical stimulation. We observe significant d

Increasing evidence indicates that anesthetics strongly affect sensory representations and network dynamics in the thalamo-cortical circuits. By contrast, processing in the early sensory systems is thought to be mostly preserved under anesthesia, based on stimulus tuning properties like those observed in wakefulness. However, in the absence of recordings tracking the activity of the same neurons across states, there is no evidence that anesthesia preserves peripheral one-to-one sensory representations. Here, tracking for the first time the same cochlear nucleus neurons across wakefulness and i

…Applying dimensionality reduction methods such as principal component analysis can improve classification performance, help discover hidden patterns, and create new features. We aimed to assemble a multi-step analysis workflow that can transform oscillatory LFP activity into a statistical representation. After filtering and downsampling the LFP signal, waveform segments for the events of i

Real world information received by sensory systems is highly dynamic. If and how the recent temporal dynamics, as in the preceding intervals, shape cortical processing is unclear. Here we recorded the cortical signals (EEG) in response to a dynamic train of tactile stimulations applied at the fingertips ( n = 64). We separated the cortical signals according to the next interval dynamics by using a joint-interval distribution (JID) spanning 100ms to 10s. This yielded a two-dimensional event-related potential (JID-ERP) at all the examined time points of the EEG signal (-200 to 300 ms latency fro

Understanding how sensory information is represented, processed, and leads to generation of complex behavior is the major goal of systems neuroscience. However, the ability to detect and manipulate such large-scale functional circuits has been hampered by the lack of appropriate tools and methods that allow for parallel recording of neuronal population activity at high spatial and temporal resolutions. We have recently demonstrated near-simultaneous recording from over 1 million neurons distributed across both hemispheres and different layers of the mouse cortex at cellular resolution. At the

…Classic dimensionality reduction methods often assume that neural representations are constrained to a fixed low-dimensional latent subspace. Consequently, this view does not capture how the latent subspace could evolve over learning, nor how high-dimensional neural activity could emerge over learning. Furthermore, the link between these empirically-observed changes in neural

Mapping behavioral actions to neural activity is a fundamental goal of neuroscience. As our ability to record large neural and behavioral data increases, there is growing interest in modeling neural dynamics during adaptive behaviors to probe neural representations (Urai et al., 2022; Krakauer et al., 2017). In particular, neural latent embeddings can reveal underlying correlates of behavior, yet, we lack non-linear techniques that can explicitly and flexibly leverage joint behavior and neural data. Here, we fill this gap with a novel contrastive learning method that jointly uses time-series d

…and it is important to know both the dimensionality and composition of the neurally relevant behavioral subspace. However, methods for data-driven identification of behavioral subspaces relevant for neural recordings is nascent. Here, we introduce external Dynamics Components Analysis (eDCA), a linear dimensionality reduction method that finds subspaces of past neural data that have maximal mutual informa

Breathing is a vital sensorimotor behavior maintained throughout life that must be concurrently robust and flexible. Despite the relative simplicity of this behavior, the neural circuits that drive and maintain breathing are distributed throughout a large population of anatomically distributed and molecularly diverse neural population that compose the rostro-caudally extended Ventral Respiratory Column (VRC) The rhythmic properties of individual VRC nuclei are well known, yet technical challenges have limited the interrogation of the entire VRC population simultaneously. Here we introduce an e

…Here, we advance the use of the dimensionality reduction approach group factor analysis (GFA), alongside la

Communication is an inherently interactive behavior involving the coordinated exchange of social signals between two or more individuals. Studies exploring the neural basis of communication in the primate brain have traditionally focused on how these signals are processed or produced, leaving many facets of vocal behavior largely unexplored. One challenge has been limitations of traditional analyses for quantifying the often variable, dynamic nature of natural communication behaviors, such as conversational exchanges. Here we recorded the activity of single neurons in prefrontal and premotor c

Learning in neural circuits manifests via slow changes in activity driving behaviour changes. Uncovering how the coordination of plasticity mechanisms found in the brain leads to such changes is a difficult problem, chiefly because we do not have access to the synaptic parameters underlying neural computation. Nevertheless, as neural population recording techniques improve and experiments track changes in neural dynamics over long durations, finer details about plastic changes could be inferred by observing and characterizing changes in neural activity. In this work, we present a step towards

Predictive modeling and dimensionality reduction of functional neuroimaging data have provided rich information about the representations and functional architectures of the human brain. While these approaches have been effective in many cases, we will discuss how neglecting the internal dynamics of the brain (e.g., spontaneous activity, global dynamics, effective connectivity) and its underlying computational principles may hinder our progress in understanding and modeling brain functions. By reexamining evidence from our previous and ongoing work, we will propose new hypotheses and direction

The understanding of the stroke lesions’ consequences is limited, and it relies mostly on behavioral reports and mere descriptive correlational information from neuroimaging techniques. Using functional magnetic resonance imaging, it has been shown that the functional synchronization between distinct regions of the brain, referred to as functional connectivity, is disrupted by stroke. Nevertheless, this kind of data is usually large and high-dimensional. Extracting useful information from the vast amount of information afforded by brain networks remains a great challenge so, in order to make s

Brains often infer low-dimensional features from high-dimensional input. For example, during visual processing a picture captured by the retina is processed hierarchically through the ventral stream, which gradually infers higher-level features encoded by the lower number of neurons. However, it is still unclear how neural networks learn to perform such tasks. Hebbian Learning (HL), a local form of plasticity that uses pre- and postsynaptic activity to change synaptic efficacy, has been used extensively in receptive field models, sparse coding, independent and principal component analysis, an

…– and is therefore amenable to dimensionality-reduction methods – standard techniques, like principal components analysis,