Topic: latent variable models

One or two two-year scholarships support research on next-generation latent-variable models for social and behavioural data. The project integrates quantitative and textual information, models changes in measurement processes over time and develops scalable inference for high-dimensional data. Scholars develop statistical theory and computational methods, collaborate through Stat4Reg and the Umeå SIMSAM Lab, and pursue independent research questions. The proposed start is 1 January 2027 or by agreement. Applications include a research plan and selected publications.

Workshop bringing together researchers working on AI for learning dynamics from data, with a special focus on information-theoretic methods. Topics include predictive information, the information bottleneck, latent-variable models, entropy and mutual-information estimation, world models of biological systems, and connections to dynamical systems and control theory. The workshop is part of the six-week NITMB program “Image-Based Scientific Machine Learning for Theories of Biological Dynamics Across Scales” (February 1–March 12, 2027). Program: AI in Biological Dynamics Scientific Focus

A one-year machine-learning research internship with the Allen Institute Neural Dynamics accelerator. The intern will analyze recordings from two genetically distinct neural populations in the striatum alongside motor output, muscle activity and behavioral video. The project asks how shared and population-specific neural activity encodes movement, using latent-variable models, regression, classification and cross-validation. Work includes reproducible Python analysis and communicating results through figures and methods that may contribute to a poster or publication. No previous neuroscience b

Characterizing the relationship between neural population activity and behavioral data is a central goal of neuroscience. While latent variable models (LVMs) are successful in describing high dimensional data, they are typically only designed for a single type of data, making it difficult to identify structure shared across different experimental data modalities. Here, we address this shortcoming by proposing an LVM which extracts shared and independent latents for distinct, simultaneously recorded experimental modalities. We do this by combining Gaussian Process Factor Analysis (GPFA), an int

Finding a low-dimensional manifold of neural signals is crucial for understanding neural computation and developing a robust brain-computer interface (BCI). Latent variable models have been proposed to find such manifolds, and proved to be successful in various cognitive and behavioral tasks. However, natural speaking is a complex coordination of multiple motor tasks, and due to such complex nature, it has been challenging to find a low-dimensional and behaviorally relevant manifold of the speech cortex. We hypothesize that contents represented on the manifold should be consistent across repea

Latent variable models have become instrumental in computational neuroscience for reasoning about neural computation. This has fostered the development of powerful offline algorithms for extracting latent neural trajectories from neural recordings. However, despite the potential of real-time algorithms to give immediate feedback to experimentalists, and enhance experimental design, they have received markedly less attention compared to their offline counterparts. We introduce the exponential family variational Kalman filter (eVKF), an online recursive Bayesian method aimed at simultaneously i

Calcium imaging (CI) has emerged as a prominent method for recording neural population activity. Therefore, CI recordings are prime candidates for the application of latent variable models designed to identify latent neural states and dynamics. However, in CI experiments spiking events are measured only indirectly via calcium ions that rapidly enter a cell during action potentials and then slowly seep back out. To analyze CI data, per-neuron fluorescence time-traces are typically either de-convolved to approximate spiking events or analyzed directly. The former approach, although enabling many

Modern datasets in neuroscience enable unprecedented inquiries into the relationship between complex behaviors and the activity of many simultaneously recorded neurons. While latent variable models can successfully extract low-dimensional embeddings from such recordings, using them to generate realistic spiking data, especially in a behavior-dependent manner, still poses a challenge. Here, we present Latent Diffusion for Neural Spiking data (LDNS), a diffusion-based generative model with a low-dimensional latent space: LDNS employs an autoencoder with structured state-space (S4) layers to pro

Modern datasets in neuroscience enable unprecedented inquiries into the relationship between complex behaviors and the activity of many simultaneously recorded neurons. While latent variable models can successfully extract low-dimensional embeddings from such recordings, using them to generate realistic spiking data, especially in a behavior-dependent manner, still poses a challenge. Here, we present Latent Diffusion for Neural Spiking data (LDNS), a diffusion-based generative model with a low-dimensional latent space: LDNS employs an autoencoder with structured state-space (S4) layers to proj

Latent variable models are widely used in neuroscience to extract dynamical structure from high-dimensional neural activity, and provide valuable insights into neural computation. However, these models are typically inferred from a single recording session and exhibit limited generalization capabilities. This is primarily due to statistical heterogeneities in neural recordings across subjects and tasks, along with lack of inductive biases for disentangling shared and dataset-specific dynamical structure. In this work, we hypothesize that similar behavioral tasks admit a corresponding family of

Multimodal data fusion can facilitate a more complete understanding of neural dynamics. In neuroscience, modalities such as neuronal spiking activity, local field potentials (LFPs), and behavioral signals capture various aspects of brain processes. By leveraging the complementary strengths of different modalities, multimodal fusion can not only construct a rich and unified representation of brain processes, but also address the limitations of single-modality analyses---such as incomplete or noisy data. An exciting recent direction has been to build models of simultaneously recorded neural and

Population recording offer a window into the underlying dynamics of brain regions. Latent Variable Models (LVMs) infer these dynamics treating the observations as projections from the latents. Evaluating the quality of these models remains challenging, as we don't have access to the true underlying dynamics. A common approach, "co-smoothing", involves jointly estimating latent variables from neural data and predicting the firing rates of held-out neurons. However, we find that optimizing this single objective often yields models that include extraneous dynamics not relevant to the underlying n