Content uploaded by Giovanni Squillero
Author content
All content in this area was uploaded by Giovanni Squillero on Jul 20, 2015
Content may be subject to copyright.
A preview of the PDF is not available
MicroGP—An Evolutionary Assembly Program Generator
Abstract and Figures
This paper describes μGP, an evolutionary approach for generating assembly programs tuned for a specific microprocessor. The approach is based on
three clearly separated blocks: an evolutionary core, an instruction library and an external evaluator. The evolutionary core
conducts adaptive population-based search. The instruction library is used to map individuals to valid assembly language programs.
The external evaluator simulates the assembly program, providing the necessary feedback to the evolutionary core. μGP has some distinctive features that allow its use in specific contexts. This paper focuses on one such context: test program
generation for design validation of microprocessors. Reported results show μGP being used to validate a complex 5-stage pipelined microprocessor. Its induced test programs outperform an exhaustive functional
test and an instruction randomizer, showing that engineers are able to automatically obtain high-quality test programs.
Figures - uploaded by Giovanni Squillero
Author content
All figure content in this area was uploaded by Giovanni Squillero
Content may be subject to copyright.
... One of its key strengths lies in its flexibility, enabling it to address a diverse range of problems in fields such as engineering, finance, and biology [81]. Moreover, GP's ability to automatically generate computer programs without human intervention saves valuable time and effort [82]. Additionally, it excels at optimizing complex functions that may be challenging for traditional methods, and its creative nature often leads to unexpected and innovative solutions, potentially uncovering new discoveries. ...
Geotechnical engineering relies heavily on predicting soil strength to ensure safe and efficient construction projects. This paper presents a study on the accurate prediction of soil strength properties, focusing on hydrated-lime activated rice husk ash (HARHA) treated soil. To achieve precise predictions, the researchers employed two grey-box machine learning models-classification and regression trees (CART) and genetic programming (GP). These models introduce innovative equations and trees that readers can readily apply to new databases. The models were trained and tested using a comprehensive laboratory database consisting of seven input parameters and three output variables. The results indicate that both the proposed CART trees and GP equations exhibited excellent predictive capabilities across all three output variables-California bearing ratio (CBR), unconfined compressive strength (UCS), and resistance value (R value) (according to the in-situ cone penetrometer test). The GP proposed equations, in particular, demonstrated a superior performance in predicting the UCS and R value parameters, while remaining comparable to CART in predicting the CBR. This research highlights the potential of integrating grey-box machine learning models with geotechnical engineering, providing valuable insights to enhance decision-making processes and safety measures in future infrastructural development projects.
Tissue Engineering and Regenerative Medicine aim to replicate and replace tissues for curing disease. However, full tissue integration and homeostasis are still far from reach. Biofabrication is an emerging field that identifies the processes required for generating biologically functional products with the desired structural organization and functionality and can potentially revolutionize the regenerative medicine domain, which aims to use patients' cells to restore the structure and function of damaged tissues and organs. However, biofabrication still has limitations in the quality of processes and products. Biofabrication processes are often improved empirically, but this is slow, costly, and provides partial results. Computational approaches can tap into biofabrication underused potential, supporting analysis, modeling, design, and optimization of biofabrication processes, speeding up their improvement towards a higher quality of products and subsequent higher clinical relevance. This work proposes a reinforcement learning-based computational design space exploration methodology to generate optimal protocols for the simulated fabrication of epithelial sheets. The optimization strategy relies on a deep reinforcement learning algorithm, the Advantage-Actor Critic, which relies on a neural network model for learning. In contrast, simulations rely on the PalaCell2D simulation framework. Validation demonstrates the proposed approach on two protocol generation targets: maximizing the final number of obtained cells and optimizing the spatial organization of the cell aggregate.Advantage-Actor Critic, which relies on a neural network model for learning. In contrast, simulations rely on the PalaCell2D simulation framework. Validation demonstrates the proposed approach on two protocol generation targets: maximizing the final number of obtained cells and optimizing the spatial organization of the cell aggregate.
This chapter describes recent advances in genetic programming of machine code. Evolutionary pro- gram induction of binary machine code is one of the fastest GP methods and the most well studied linear approach. The technique has previously been known as Compiling Genetic Programming System (CGPS) but to avoid confusion with methods using an actual compiler and to separate the system from the method, the name has been changed to Automatic Induction of Machine Code with Genetic Programming (AIM-GP). AIM-GP stores individuals as a linear string of native binary ma- chine code, which is directly executed by the processor. The absence of an interpreter and complex memory handling allows increased speed of several orders of magnitudes. AIM-GP has so far been applied to processors with a fixed instruction length (RISC) using integer arithmetics. This chapter describes several new advances to the AIM-GP method which are important for the applicability of the technique. Such advances include enabling the induction of code for CISC processors such as the most widespread computer architecture INTEL x86 as well as JAVA and many embedded processors. The new technique also makes AIM-GP more portable in general and simplifies the adaptation to any processor architecture. Other additions include the use of floating point instructions, control flow in- structions, ADFs and new genetic operators e.g. aligned homologous crossover. We also discuss the benefits and drawbacks of register machine GP versus tree-based GP. This chapter is meant to be a directed towards the practitioner, who wants to extend AIM-GP to new architectures and application domains.
An effort is made to improve the performance of the learning machine described in Part I, and the over-all effect of various changes is considered. Comparative runs by machines without the scoring mechanism indicate that the grading of individual instructions can aid in the learning process. A related study is made in which automatic debugging of programs is taken as a special case of machine search. The ability to partition problems and to deal with parts in order of difficulty proves helpful.
A new signature-table technique is described together with an improved book-learning procedure which is thought to be much superior to the linear polynomial method. Full use is made of the so-called “alpha-beta” pruning and several forms of forward pruning to restrict the spread of the move tree and to permit the program to look ahead to a much greater depth than it otherwise could do. While still unable to outplay checker masters, the program's playing ability has been greatly improved.tplay checker masters, the
Machines would be more useful if they could learn to perform tasks for which they were not given precise methods. Difficulties that attend giving a machine this ability are discussed. It is proposed that the program of a stored-program computer be gradually improved by a learning procedure which tries many programs and chooses, from the instructions that may occupy a given location, the one most often associated with a successful result. An experimental test of this principle is described in detail. Preliminary results, which show limited success, are reported and interpreted. Further results and conclusions will appear in the second part of the paper.
ABSTRACT Developing a new leading - edge Intel Architecture microprocessor is an immensely complicated undertaking The microarchitecture of the Pentium ò 4 processor is significantly more complex than any previous Intel Architecture microprocessor, so the challenge of validating the logical correctness of the design in a timely fashion was indeed a daunting one In order to meet this challenge, we applied a number of innovative tools and methodologies, which enabled us to keep functional validation off the critical path to tapeout while meeting our goal of ensuring that first silicon was functional enough to boot operating systems and run applications This in turn enabled the post - silicon validation teams to quickly "peel the onion", resulting in an elapsed time of only ten months from initial tapeout to production shipment qualification, an Intel record for a new IA - microarchitecture This paper describes how we went about the task of validating the Pentium 4 processor and what we found along the way We hope that other microprocessor designers and validators will be able to benefit from our experience and insights As Doug Clark has remarked "Finding a bug should be a cause for celebration Each discovery is a small victory; each marks an incremental improvement in the design " [1]