Omid Mohamad Nezami | Macquarie University (original) (raw)

Papers by Omid Mohamad Nezami

Lecture Notes in Computer Science, 2010

یکی از موضوعات نسبتاً جدید در حوزۀ پردازش زبان طبیعی، ساخت سیستم‌های تولید خودکار زبان است. این س... more یکی از موضوعات نسبتاً جدید در حوزۀ پردازش زبان طبیعی، ساخت سیستم‌های تولید خودکار زبان است. این سیستم‌ها، با بکارگیری شیوه‌های رایج در هوش مصنوعی و زبان‌شناسی رایانه‌ای اقدام به تولید خودکار متنِ قابل‌فهم به زبان‌های گوناگون می‌نمایند. متن تولیدشده ممکن است گزارش، نامه، توضیح، خلاصه، مقاله، پیام، داستان و غیره باشد.
در این مقاله، قصد داریم سیستمی برای تولید خودکار نامه‌های اداری فارسی ارائه کنیم. این سیستم قادر به تولید سه نوع نامۀ اداری شامل دعوتنامه، تبریکات و تقدیرنامه است و دو حالتِ تولید متن بصورت پیش‌فرض و تولید متن با استفاده از اطلاعات شخصی کاربر را دربرمی‌گیرد. خروجی این سیستم به‌گونه‌ای است که نیاز به دخالت انسان برای ویرایش متن تولیدشده ندارد. در طراحی این سیستم، از روش مبتنی بر داده استفاده شده است. سه روشِ مختلف که همگی از مدل احتمالاتی برای پیش‌بینیِ کلمه یا کلمات بعدی استفاده می‌کنند آموزش داده شدند. این سه روش شامل استفاده از احتمال مدل زبانی 4-گرام کلمات، احتمال دنبالۀ 4-تایی از کلمات و تركيب احتمالات مدل‌زبانی و دنبالۀ 3-تایی از کلمات هستند. نامه‌های تولیدشده با استفاده از معیار بلو (BLEU) و امتیازدهیِ انسانی ارزیابی شدند که بهترین نتیجه مربوط به روش اول با امتیازِ بلوی 85/0 بوده است.

The problem of early convergence in the Particle Swarm Optimization (PSO) algorithm often causes ... more The problem of early convergence in the Particle Swarm Optimization (PSO) algorithm often causes the search process to be
trapped in a local optimum. This problem often occurs when the diversity of the swarm decreases and the swarm cannot escape
from a local optimal. In this paper, a novel dynamic diversity enhancement particle swarm optimization (DDEPSO) algorithm is
introduced. In this variant of PSO, we periodically replace some of the swarm's particles by artificial ones, which are generated
based on the history of the search process, in order to enhance the diversity of the swarm and promote the exploration ability of
the algorithm. Afterwards, we update the velocity of the artificial particles in corresponding generating period by a new velocity
equation with the minimum inertia weight in order to enhance the exploitation potentiality of the swarm. The performance of this
approach has been tested on the set of twelve standard unimodal and multimodal (Rotated or unrotated) benchmark problems and
the results have been compared with our previous work as well as four other variants of the PSO algorithm. The numerical results
demonstrate that the proposed algorithm outperforms others in most of the test cases taken in this study.

Diversity control in the particle swarm optimization (PSO) algorithm is one of the important iss... more Diversity control in the particle swarm
optimization (PSO) algorithm is one of the important issues
that influence the process of finding global optimal solution.
In this study we create a historical process to find best area
of the search space for population dispersion guide on PSO
algorithm, and name Diversity Guided Particle Swarm
Optimization algorithm (DGPSO) algorithm. Hence we
propose a mechanism to guide the swarm based on diversity
by using a diversifying process in order to detect suitable
positions of the search space (points with fairly good fitness,
and good distance from current distribution of the swarm
particles) to disperse or relocating some of existing particles,
hoped to increase diversity level of the swarm and escape
from local optimal by detecting better area of the search
space. This model uses a diversity measuring, and swarm
dispersion mechanism to control the evolutionary process
alternating between exploring and exploiting behavior. The
numerical results show that the proposed algorithm
outperforms other algorithms in most of the test cases taken
in this study.

Speed of convergence in the PSO is very high, and this issue causes to the algorithm can't invest... more Speed of convergence in the PSO is very high, and this issue causes to the algorithm can't investigate search space truly, When diversity of the population decreasing, all the population start to liken together and the algorithm converges to local optimal swiftly. In this paper we implement a new idea for better control of the diversity and have a good control of the algorithm's behavior between exploration and exploitations phenomena to preventing premature convergence. In our approach we have control on diversity with generating diversified artificial particles (DAP) and injection them to the population by a particular mechanism when diversity lessening, named Particle Swarm Optimization algorithm based on Diversified Artificial Particles (PSO-DAP). The performance of this approach has been tested on the set of ten standard benchmark problems and the results are compared with the original PSO algorithm in two models, Local ring and Global star topology. The numerical results show that the proposed algorithm outperforms the basic PSO algorithms in all the test cases taken in this study

Controlling a biped robot with a high degree of freedom to achieve stable and straight movement p... more Controlling a biped robot with a high degree of freedom to achieve stable and straight movement patterns is a complex problem. With growing computational power of computer hardware, high resolution real time simulation of such robot models has become more and more applicable. This paper presents a novel approach to generate bipedal gait for humanoid locomotion. This approach is based on modified Truncated Fourier Series (TFS) for generating angular trajectories. It is also the first time that Particle Swarm Optimization (PSO) is used to find the best angular trajectory and optimize TFS. This method has been implemented on Simulated NAO robot in Robocup 3D soccer simulation environment (rcssserver3d). To overcome inherent noise of the simulator we applied a Resampling algorithm which could lead the robustness in nondeterministic environments. Experimental results show that PSO optimizes TFS faster and better than GA to generate straighter and faster humanoid locomotion.

This paper describes the main research focus of Mechatronic Research Laboratory (MRL) Mixed Rea... more This paper describes the main research focus of
Mechatronic Research Laboratory (MRL) Mixed Reality team
entering the LARC 2010 competitions. MRL's activities
including software development, artificial intelligence and
electronic achievements are described in this paper.

This paper describes the design and analysis a simple bipedal locomotion algorithm based on Fouri... more This paper describes the design and analysis a simple bipedal locomotion algorithm based on Fourier series as a gait generator. The algorithm uses a Truncated Fourier Series (TFS) to generate control signal for the bipedal robot. This gait generator has only 7 parameters to generate all walking trajectory. Genetic Algorithm (GA) as parameters searching method with Resampling technique is used to learn the robot how to walk robust and straightly in noisy environment. This approach is tested on a Simulated NAO robot in Robocup soccer simulation environment. Simulation results show the proposed TFS with can generate smooth and continuous walking pattern and robot can walk robustly and straightly.

This paper briefly describes efforts of MRL 3D Soccer Simulation Team during past ye... more This paper briefly describes efforts of MRL 3D Soccer Simulation Team during past year to develop humanoid locomotion skills for simulated robot in this league.Different approaches are followed and implemented in this
way and Evolutionary Algorithms are selected among them as an effective solution to overcome complexity of implementation of such skills. However, the model that EA must work on is too important which should guarantee the convergence of learning. Two implemented models are described and
experimental results are explained. Examining more machine learning techniques in this field, MRL is going to achieve a method which is fast and optimum to implement different humanoid locomotion skills and is discussed in this paper.

Awards by Omid Mohamad Nezami

2010 Singapore, Singapore

Projects by Omid Mohamad Nezami

Lecture Notes in Computer Science, 2010