This paper considers the fundamental problem of learning a complete (orthogonal) dictionary from samples of sparsely generated signals. Most existing methods solve the dictionary (and sparse representations) based on heuristic algorithms, usually without theoretical guarantees for either optimality or complexity. The recent $\ell^1$-minimization based methods do provide such guarantees but the associated algorithms recover the dictionary one column at a time. In this work, we propose a new formulation that maximizes the $\ell^4$-norm over the orthogonal group, to learn the entire dictionary. We prove that under a random data model, with nearly minimum sample complexity, the global optima of the $\ell^4$ norm are very close to signed permutations of the ground truth. Inspired by this observation, we give a conceptually simple and yet effective algorithm based on "matching, stretching, and projection" (MSP). The algorithm provably converges locally at a superlinear (cubic) rate and cost per iteration is merely an SVD. In addition to strong theoretical guarantees, experiments show that the new algorithm is significantly more efficient and effective than existing methods, including KSVD and $\ell^1$-based methods. Preliminary experimental results on mixed real imagery data clearly demonstrate advantages of so learned dictionary over classic PCA bases.
翻译:本文考虑了从微弱生成信号的样本中学习完整( orthogonal) 字典的根本问题。 多数现有方法基于超自然算法解决了字典( 和稀少表示), 通常没有优化或复杂程度的理论保障。 最近以$ell\ $1$- $- $- 最小化为基础的方法确实提供了这样的保障, 但相关的算法一次恢复了字典的一栏。 在这项工作中, 我们提出了一个新的公式, 使美元/ 4$- norm 最大化于正单数组, 以学习整个字典。 我们证明, 在随机数据模型下, 几乎是最低的样本复杂性, $/ $/ 4$/ 标准的全球选择非常接近于签名的地面真相配置。 受此观察的启发, 我们给出了一个基于“ 匹配、 延伸和 预测” ( MSP ) 的概念简单而有效的算法。 我们提出的算法在超级线性( ) 率和成本中, 仅仅是一个 SVD 。 除了强有力的理论保证外, 实验表明新的算法比 真正的K- drialimalalimageleglevelopalal1 的模型更明显有效, 方法要高出。