Program Synthesis with Priority Queue Training
نویسندگان
چکیده
We consider the task of program synthesis in the presence of a reward function over the output of programs, where the goal is to find programs with maximal rewards. We introduce an iterative optimization scheme, where we train an RNN on a dataset of K best programs from a priority queue of the generated programs so far. Then, we synthesize new programs and add them to the priority queue by sampling from the RNN. We benchmark our algorithm, called priority queue training (or PQT), against genetic algorithm and reinforcement learning baselines on a simple but expressive Turing complete programming language called BF. Our experimental results show that our simple PQT algorithm significantly outperforms the baselines. By adding a program length penalty to the reward function, we are able to synthesize short, human readable programs.
منابع مشابه
Neural Program Synthesis with Priority Queue Training
We consider the task of program synthesis in the presence of a reward function over the output of programs, where the goal is to find programs with maximal rewards. We employ an iterative optimization scheme, where we train an RNN on a dataset of K best programs from a priority queue of the generated programs so far. Then, we synthesize new programs and add them to the priority queue by samplin...
متن کاملA Finite Capacity Priority Queue with Discouragement
In this paper we report on a study of a two level preemptive priority queue with balking and reneging for lower priority level. The inter-arrival and the service times for both levels follow exponential distribution. We use a finite difference equation approach for solving the balance equations of the governing queuing model whose states are described by functions of one independent variable. H...
متن کاملOn Finite Queue with Two Types of Failures and Preemptive Priority (RESEARCH NOTE)
We study the single server queueing system with two types of Abstract failure to servicechannels including the preemptive priority to the repair of major failure. The units arrive at thesystem in a poisson fashion and are served exponenitally. The steady - state probabilities ofvarious states by using generating function have been obtained.
متن کاملPriority-Driven Acoustic Modeling for Virtual Environments
Geometric acoustic modeling systems spatialize sounds according to reverberation paths from a sound source to a receiver to give an auditory impression of a virtual 3D environment. These systems are useful for concert hall design, teleconferencing, training and simulation, and interactive virtual environments. In many cases, such as in an interactive walkthrough program, the reverberation paths...
متن کاملA New Form of Efficient Tree-based Priority Queues for Discrete Event Simulation
A priority queue plays an important role in stochastic discrete event simulations for as much as 40% of a simulation execution time is consumed by the pending event set management. This article describes a new form of tree-based priority queues which employs the demarcation procedure to systematically split a single tree-based priority queue into many smaller trees, each divided by a logical ti...
متن کامل