Chethan Pandarinath

This paper introduces DynISys, a model that infers neural dynamics and communication from multi-area recordings, revealing mechanisms for motor planning and execution.

This paper introduces a multi-user transformer-based decoder for intracortical speech BCIs, achieving lower word error rates with less training data than individual models.

This paper tests deep neural ensembles in real-time speech BCIs, showing reduced word error rates and introducing a computationally efficient pseudoensembling method.

This study shows that meta-learning in rats involves changes in prefrontal cortical dynamics that enable rule-based reward learning.

This study decodes speech mode and loudness from ventral precentral gyrus activity in two participants, achieving high accuracy in predicting speech parameters and developing a closed-loop decoder for expressive speech neuroprostheses.

This paper demonstrates cross-brain transfer learning for intracortical BCIs decoding speech and handwriting, showing improved performance with limited training data and specific decoder design.

EMUsort improves MUAP spike sorting accuracy by addressing challenges like long delays, complex waveforms, and high overlap in motor unit recordings.

The paper uses artificial intelligence to analyze spinal interneuron activity and muscle output in cats, revealing how state space trajectories capture timing and magnitude variations in muscle activation across steps.

This study investigates how preparatory activity in the human motor cortex encodes movement features like direction, curvature, and distance, and demonstrates its use for predictive brain-computer interface control.

The paper uses MoSeq to analyze behavioral changes in dystonia mouse models after trihexyphenidyl treatment, finding genotype and sex-specific shifts in behavioral structure.

The paper shows that neural activity in human premotor cortex combines finger movements through pseudo-linear summation, with deviations that improve decoding performance.

This paper introduces NoMAD, a method that stabilizes brain-computer interfaces by aligning latent dynamics, achieving stable decoding without supervised recalibration over long periods.

The paper introduces CtDToolkit, a framework providing synthetic datasets and metrics to evaluate and improve neural dynamics models of neural computation.

The paper studies neural encoding of inner speech in motor cortex and shows it can be decoded, but also identifies a neural dimension that can prevent unintentional decoding of private thoughts.

The paper maps human motor cortex using microelectrode recordings, finding whole-body representations and challenging the traditional motor homunculus model.

BRAND provides a platform for closed-loop experiments with deep networks, achieving low latency and supporting multiple programming languages.

The paper introduces FALCON, a benchmark suite for evaluating robustness of iBCI decoders using standardized datasets and evaluation metrics.

The paper introduces Myomatrix arrays, a high-resolution electrode technology for recording muscle activity from motor units across species during natural behaviors.

This paper introduces AutoLFADS, an automated framework for estimating neural population dynamics from neural data without supervisory information, validated on multiple datasets.

This paper introduces RADICaL, a deep learning framework that infers single-trial neural population dynamics from calcium imaging with sub-frame temporal resolution, showing improved accuracy in synthetic and real mouse data.

This paper uses a deep learning model to estimate muscle activation from EMG signals, showing improved performance over traditional methods in predicting joint kinematics and force.

The paper introduces a non-recurrent model, Neural Data Transformer, that captures neural population dynamics as effectively as recurrent models but with faster inference.

This study shows that finger movement and force have distinct neural representations in motor and premotor cortices, using electrocorticography and deep learning methods.

This paper introduces LFADS, a deep learning method to infer single-trial neural population dynamics from spiking data, demonstrating improved accuracy in predicting behavior and extracting dynamics across datasets.