Polyhedral-Based Compilation Framework for In-Memory Neural Network Accelerators

Memristor-based processing-in-memory architecture is a promising solution to the memory bottleneck in the neural network ( NN ) processing. A major challenge for the programmability of such architectures is the automatic compilation of high-level NN workloads, from various operators to the memristor-based hardware that may provide programming interfaces with different granularities. This article proposes a source-to-source compilation framework for such memristor-based NN accelerators, which can conduct automatic detection and mapping of multiple NN operators based on the flexible and rich representation capability of the polyhedral model. In contrast to previous studies, it implements support for pipeline generation to exploit the parallelism in the NN loads to leverage hardware resources for higher efficiency. The evaluation based on synthetic kernels and NN benchmarks demonstrates that the proposed framework can reliably detect and map the target operators. Case studies on typical memristor-based architectures also show its generality over various architectural designs. The evaluation further demonstrates that compared with existing polyhedral-based compilation frameworks that do not support the pipelined execution, the performance can upgrade by an order of magnitude with the pipelined execution, which emphasizes the necessity of our improvement.

Integrated Social Value at Universities: A Guarantee for Public Subsidies

The objective of this work was to monetize the integrated social value generated by a university during an academic year to justify the use of public grants received and determine their relationship with the institution’s stakeholders. Monetization allows us to analyze the efficiency in the use of received subsidies. The methodology used herein was the polyhedral model, which was applied to a university during the 2017–2018 academic year in order to reflect the financial and social accounting of the institution, its relationship with stakeholders, and its efficiency in economic and social management. The results of the study highlighted the monetization of the activity of the UVic-UCC over the course of 2017–2018, guaranteeing the economic and social efficiency and dynamization of the University, as well as information regarding the return to the Public Administration by the institution and the good use of the grants it received. The monetization of economic transactions and the value variables of stakeholders allowed us to assess the impact of the institution and the added value it created. The findings indicated that the integrated social value of the UVic-UCC in the 2017–2018 academic year was EUR 64,268,260, and this guaranteed the good economic and social management of the institution and the good use of the grants it received.

Statistical Simulation of the Probability Distribution Kernel and Its Application to Solve the Quantile Optimization Problem with the Bilinear Loss Function

The article considers a plane quantile optimization problem with a bilinear loss function, which, using suffi cient optimality conditions, is reduced to a linear programming problem. The reduction is based on the use of a polyhedral model of the kernel of the probability distribution of the vector of random parameters. To build this model, an algorithm based on the method of statistical modeling is proposed. A description of the software package for constructing a kernel model for a number of probability distributions of random parameters is given.