Abstract: The present disclosure relates to a discovery platform including machine-learning techniques for using medical imaging data to study a phenotype of interest, such as complex diseases with weak or unknown genetic drivers. An exemplary method of identifying a covariant of interest with respect to a phenotype comprises: receiving covariant information of a covariate class and corresponding phenotypic image data related to the phenotype obtained from a group of clinical subjects; inputting the phenotypic image data into a trained unsupervised machine-learning model to obtain a plurality of embeddings in a latent space, each embedding corresponding to a phenotypic state reflected in the phenotypic image data; and determining, based on the covariant information for the group of clinical subjects, the plurality of embeddings, and one or more linear regression models, an association between each candidate covariant of a plurality of candidate covariants and the phenotype state to identify the covariant of interest.

Abstract: Examples of the present disclosure describe systems and methods for probabilistic neural network architecture generation. In an example, an underlying distribution over neural network architectures based on various parameters is sampled using probabilistic modeling. Training data is evaluated in order to iteratively update the underlying distribution, thereby generating a probability distribution over the neural network architectures. The distribution is iteratively trained until the parameters associated with the neural network architecture converge. Once it is determined that the parameters have converged, the resulting probability distribution may be used to generate a resulting neural network architecture. As a result, intermediate architectures need not be fully trained, which dramatically reduces memory usage and/or processing time.

Abstract: Examples of the present disclosure describe systems and methods for probabilistic neural network architecture generation. In an example, an underlying distribution over neural network architectures based on various parameters is sampled using probabilistic modeling. Training data is evaluated in order to iteratively update the underlying distribution, thereby generating a probability distribution over the neural network architectures. The distribution is iteratively trained until the parameters associated with the neural network architecture converge. Once it is determined that the parameters have converged, the resulting probability distribution may be used to generate a resulting neural network architecture. As a result, intermediate architectures need not be fully trained, which dramatically reduces memory usage and/or processing time.

Abstract: Embodiments of the disclosure include implementing a ML-enabled cellular disease model for validating an intervention, identifying patient populations that are likely responders to an intervention, and developing a therapeutic structure-activity relationship screen. To generate a cellular disease model, data is combined from human genetic cohorts, from the literature, and from general-purpose cellular or tissue-level genomic data to unravel the set of factors (e.g., genetic, environmental, cellular factors) that give rise to a particular disease. In vitro cells are engineered using the set of factors to generate training data for training machine learning models that are useful for implementing cellular disease models.

Abstract: Examples of the present disclosure describe systems and methods for probabilistic neural network architecture generation. In an example, an underlying distribution over neural network architectures based on various parameters is sampled using probabilistic modeling. Training data is evaluated in order to iteratively update the underlying distribution, thereby generating a probability distribution over the neural network architectures. The distribution is iteratively trained until the parameters associated with the neural network architecture converge. Once it is determined that the parameters have converged, the resulting probability distribution may be used to generate a resulting neural network architecture. As a result, intermediate architectures need not be fully trained, which dramatically reduces memory usage and/or processing time.

Abstract: Systems and methods for selecting a neural network for a machine learning problem are disclosed. A method includes accessing an input matrix. The method includes accessing a machine learning problem space associated with a machine learning problem and multiple untrained candidate neural networks for solving the machine learning problem. The method includes computing, for each untrained candidate neural network, at least one expressivity measure capturing an expressivity of the candidate neural network with respect to the machine learning problem. The method includes computing, for each untrained candidate neural network, at least one trainability measure capturing a trainability of the candidate neural network with respect to the machine learning problem. The method includes selecting, based on the at least one expressivity measure and the at least one trainability measure, at least one candidate neural network for solving the machine learning problem.