Abstract: A computer-implemented method for a machine learning (ML) system includes receiving a first image frame and a second frame from a sensor, wherein the first and second image frames are time series data, determining a first flow state and a first latent state of the first image frame, determining a Deep Equilibrium Model (DEQ) based fix point solution via a root finding method based on the first flow state, the first latent state, and a layer function to obtain an estimated flow and latent state, receiving a third image frame, wherein the second and third image frames are time series data, determining the fix point solution via the root finding method based on the estimated flow state, the estimated latent state, and layer function to obtain an updated flow state and updated latent state, and outputting the updated flow state.

Abstract: Systems and methods for operating a deep equilibrium (DEQ) model in a neural network are disclosed. DEQs solve for a fixed point of a single nonlinear layer, which enables decoupling the internal structure of the layer from how the fixed point is actually computed. This disclosure discloses that such decoupling can be exploited while substantially enhancing this fixed point computation using a custom neural solver.

Abstract: A computer-implemented method for a classification and training a neural network includes receiving input at the neural network, wherein the input includes a plurality of resolution inputs of varying resolutions, outputting a plurality of feature tensors for each corresponding resolution of the plurality of resolution inputs, fusing the plurality of feature tensors utilizing upsampling or down sampling for the vary resolutions, utilizing an equilibrium solver to identify one or more prediction vectors from the plurality of feature tensors, and outputting a loss in response to the one or more prediction vectors.

Abstract: Regularized training of a Deep Equilibrium Model (DEQ) is provided. A regularization term is computed using a predefined quantity of random samples and the Jacobian matrix of the DEQ, the regularization term penalizing the spectral radius of the Jacobian matrix. The regularization term is included in an original loss function of the DEQ to form a regularized loss function. A gradient of the regularized loss function is computed with respect to model parameters of the DEQ. The gradient is used to update the model parameters.

Abstract: A computer-implemented method for a classification and training a neural network includes receiving input at the neural network, wherein the input includes a plurality of resolution inputs of varying resolutions, outputting a plurality of feature tensors for each corresponding resolution of the plurality of resolution inputs, fusing the plurality of feature tensors utilizing upsampling or down sampling for the vary resolutions, utilizing an equilibrium solver to identify one or more prediction vectors from the plurality of feature tensors, and outputting a loss in response to the one or more prediction vectors.

Abstract: A simulation includes converting a molecular dynamics snapshot of elements within a multi-element system into a graph with atoms as nodes of the graph; defining a matrix such that each column of the matrix represents a node in the graph; defining a distance matrix according to a set of relative positions of each of the atoms; iterating through the GTFF using an attention mechanism, operating on the matrix and augmented by incorporating the distance matrix, to pass hidden state from a current layer of the GTFF to a next layer of the GTFF; performing a combination over the columns of the matrix to produce a scalar molecular energy; making a backward pass through the GTFF, iteratively calculating derivatives at each of the layers of the GTFF to compute a prediction of force acting on each atom; and returning the prediction of the force acting on each atom.

Abstract: A simulation includes converting a molecular dynamics snapshot of elements within a multi-element system into a graph with atoms as nodes of the graph; defining a matrix such that each column of the matrix represents a node in the graph; defining a distance matrix according to a set of relative positions of each of the atoms; iterating through the GTFF using an attention mechanism, operating on the matrix and augmented by incorporating the distance matrix, to pass hidden state from a current layer of the GTFF to a next layer of the GTFF; performing a combination over the columns of the matrix to produce a scalar molecular energy; making a backward pass through the GTFF, iteratively calculating derivatives at each of the layers of the GTFF to compute a prediction of force acting on each atom; and returning the prediction of the force acting on each atom.

Abstract: Some embodiments are directed to a neural network comprising an iterative function (z[i+1]=ƒ(z[i], ?, c(?)). Such an iterative function is known in the field of machine learning to be representable by a stack of layers which have mutually shared weights. According to some embodiments the stack of layers may during training be replaced by the use of a numerical root-finding algorithm to find an equilibrium of the iterative function in which a further execution of the iterative function would not substantially further change the output of the iterative function. Effectively, the stack of layers may be replaced by a numerical equilibrium solver. The use of the numerical root-finding algorithm is demonstrated to greatly reduce the memory footprint during training while achieving similar accuracy as state-of-the-art prior art models.