Search for: All records

Coded distributed computation has become common practice for performing gradient descent on large datasets to mitigate stragglers and other faults. This paper proposes a novel algorithm that encodes the partial derivatives themselves and furthermore optimizes the codes by performing lossy compression on the derivative codewords by maximizing the information contained in the codewords while minimizing the information between the codewords. The utility of this application of coding theory is a geometrical consequence of the observed fact in optimization research that noise is tolerable, sometimes even helpful, in gradient descent based learning algorithms since it helps avoid overfitting and local minima. This stands in contrast with much current conventional work on distributed coded computation which focuses on recovering all of the data from the workers. A second further contribution is that the low-weight nature of the coding scheme allows for asynchronous gradient updates since the code can be iteratively decoded; i.e., a worker’s task can immediately be updated into the larger gradient. The directional derivative is always a linear function of the direction vectors; thus, our framework is robust since it can apply linear coding techniques to general machine learning frameworks such as deep neural networks.

Abstract On single-cell RNA-sequencing data, we consider the problem of assigning cells to known cell types, assuming that the identities of cell-type-specific marker genes are given but their exact expression levels are unavailable, that is, without using a reference dataset. Based on an observation that the expected over-expression of marker genes is often absent in a nonnegligible proportion of cells, we develop a method called scSorter. scSorter allows marker genes to express at a low level and borrows information from the expression of non-marker genes. On both simulated and real data, scSorter shows much higher power compared to existing methods.

Photoelectron spectroscopy and quantum chemistry studies are used to investigate the structure and bonding of AuB 8 − . Global minimum sturctural searches show that AuB 8 − possesses a chair-like structure, which can be viewed as Au + bonded to the edge of the doubly-aromatic B 8 2− borozene, Au + [η 2 -B 8 2− ]. Chemical bonding analyses reveal that the AuB 8 − is a novel borozene complex with unique Au–borozene bonding.

Layer-resolved state preparation and imaging is reported in highly tunable two-dimensional layers of ultracold potassium–rubidium molecules in an optical lattice.