Search for: All records

Sea Island cotton ( Gossypium barbadense ) is world-renowned for its superior natural fiber. Although fiber strength is one of the most important fiber quality traits, genes contributing to fiber strength are poorly understood. Production of sea island cotton also is inextricably linked to improving its relatively low yield, thus enhancing the importance of joint improvement of both fiber quality and yield. We used genomic variation to uncover the genetic evidence of trait improvement resulting from pedigree breeding of Sea Island cotton. This pedigree was aimed at improving fiber strength and yielded an elite cultivar, XH35. Using a combination of genome-wide association study (GWAS) and selection screens, we detected 82 putative fiber-strength-related genes. Expression analysis confirmed a calmodulin-like gene, GbCML7 , which enhanced fiber strength in a specific haplotype. This gene is a major-effect gene, which interacts with a minor-effect gene, GbTUA3 , facilitating the enhancement of fiber strength in a synergistic fashion. Moreover, GbCML7 participates in the cooperative improvement of fiber strength, fiber length, and fiber uniformity, though a slight compromise exists between the first two of these traits and the latter. Importantly, GbCML7 is shown to boost yield in some backgrounds by increasing multiple yield components to varying degrees, especially boll number. Our work provides valuable genomic evidence and a key genetic factor for the joint improvement of fiber quality and yield in Sea Island cotton. 

We propose methods for distributed graph-based multi-task learning that are based on weighted averaging of messages from other machines. Uniform averaging or diminishing stepsize in these methods would yield consensus (single task) learning. We show how simply skewing the averaging weights or controlling the stepsize allows learning different, but related, tasks on the different machines. 

We study the stochastic optimization of canonical correlation analysis (CCA), whose objective is nonconvex and does not decouple over training samples. Although several stochastic gradient based optimization algorithms have been recently proposed to solve this problem, no global convergence guarantee was provided by any of them. Inspired by the alternating least squares/power iterations formulation of CCA, and the shift-and-invert preconditioning method for PCA, we propose two globally convergent meta-algorithms for CCA, both of which transform the original problem into sequences of least squares problems that need only be solved approximately. We instantiate the meta-algorithms with state-of-the-art SGD methods and obtain time complexities that significantly improve upon that of previous work. Experimental results demonstrate their superior performance.