Zheng Zhao

Jiaqi Gu, Chenghao Feng, Zheng Zhao et al.

Optical neural networks (ONNs) have demonstrated record-breaking potential in high-performance neuromorphic computing due to their ultra-high execution speed and low energy consumption. However, current learning protocols fail to provide scalable and efficient solutions to photonic circuit optimization in practical applications. In this work, we propose a novel on-chip learning framework to release the full potential of ONNs for power-efficient in situ training. Instead of deploying implementation-costly back-propagation, we directly optimize the device configurations with computation budgets and power constraints. We are the first to model the ONN on-chip learning as a resource-constrained stochastic noisy zeroth-order optimization problem, and propose a novel mixed-training strategy with two-level sparsity and power-aware dynamic pruning to offer a scalable on-chip training solution in practical ONN deployment. Compared with previous methods, we are the first to optimize over 2,500 optical components on chip. We can achieve much better optimization stability, 3.7x-7.6x higher efficiency, and save >90% power under practical device variations and thermal crosstalk.

Jun Liu, Zheng Zhao, Ruiwen Zhang

Coordinate descent is one of the most popular approaches for solving Lasso and its extensions due to its simplicity and efficiency. When applying coordinate descent to solving Lasso, we update one coordinate at a time while fixing the remaining coordinates. Such an update, which is usually easy to compute, greedily decreases the objective function value. In this paper, we aim to improve its computational efficiency by reducing the number of coordinate descent iterations. To this end, we propose a novel technique called Successive Ray Refinement (SRR). SRR makes use of the following ray continuation property on the successive iterations: for a particular coordinate, the value obtained in the next iteration almost always lies on a ray that starts at its previous iteration and passes through the current iteration. Motivated by this ray-continuation property, we propose that coordinate descent be performed not directly on the previous iteration but on a refined search point that has the following properties: on one hand, it lies on a ray that starts at a history solution and passes through the previous iteration, and on the other hand, it achieves the minimum objective function value among all the points on the ray. We propose two schemes for defining the search point and show that the refined search point can be efficiently obtained. Empirical results for real and synthetic data sets show that the proposed SRR can significantly reduce the number of coordinate descent iterations, especially for small Lasso regularization parameters.

Zheng Zhao, Jun Liu

Screening is an effective technique for speeding up the training process of a sparse learning model by removing the features that are guaranteed to be inactive the process. In this paper, we present a efficient screening technique for sparse support vector machine based on variational inequality. The technique is both efficient and safe.

Jun Liu, Zheng Zhao, Jie Wang et al.

Sparse learning techniques have been routinely used for feature selection as the resulting model usually has a small number of non-zero entries. Safe screening, which eliminates the features that are guaranteed to have zero coefficients for a certain value of the regularization parameter, is a technique for improving the computational efficiency. Safe screening is gaining increasing attention since 1) solving sparse learning formulations usually has a high computational cost especially when the number of features is large and 2) one needs to try several regularization parameters to select a suitable model. In this paper, we propose an approach called "Sasvi" (Safe screening with variational inequalities). Sasvi makes use of the variational inequality that provides the sufficient and necessary optimality condition for the dual problem. Several existing approaches for Lasso screening can be casted as relaxed versions of the proposed Sasvi, thus Sasvi provides a stronger safe screening rule. We further study the monotone properties of Sasvi for Lasso, based on which a sure removal regularization parameter can be identified for each feature. Experimental results on both synthetic and real data sets are reported to demonstrate the effectiveness of the proposed Sasvi for Lasso screening.