Overview / Usage

The proposed project aims to provide students with practical experience and a comprehensive understanding of the Intel AI portfolio. Through a series of assignments and workshops, students will have the opportunity to explore the benefits of utilizing Intel AI software and hardware co-optimized solutions.

The strategy for this project involves designing hands-on assignments and workshops that heavily utilize Intel AI software, DevCloud, supercomputing resources, and local resources such as notebooks. By providing access to these resources, students can gain practical experience and actively engage with the materials.

The assignments and workshops will cover various topics, including Perceptron, Neural Networks, and Search algorithms. Students will learn about the Perceptron node, different Perceptron topologies (including augmented vectors), training algorithms, and the convergence theorem. They will also delve into topics like Multilayer Perceptron network architecture, the Backpropagation algorithm, Genetic Algorithms, Bayesian classification, Ant Colony Optimization algorithm, and optimization exercises using Intel SigOpt.

Throughout the project, the expected deliverables include a repository containing the project materials, assignments, workshops content, hands-on exercises, and their resolutions. The duration of the project is set to be 10 months, starting from March 2023 and concluding in December 2023.

By engaging in this project, students will acquire practical skills in utilizing Intel AI software and hardware solutions. They will gain hands-on experience in implementing various AI algorithms and optimization techniques using Intel Python, Scikit-learn extensions, and PyTorch. The knowledge and experience gained through this project can be valuable for students in their future endeavors, whether in academia or industry, as AI continues to play a significant role in various domains.

Methodology / Approach

In this project, we employ a technology-driven approach to solve the problem at hand. Our development process involves the utilization of various frameworks, standards, and techniques to ensure efficient and effective learning and implementation.

The core technology used in this project is Python, a versatile and popular programming language known for its simplicity and extensive libraries. In particular, we make use of Intel® Distribution for Python, which provides optimized performance for AI-related tasks on Intel architectures. Additionally, we leverage the capabilities offered by Intel® oneAPI Toolkits, which enable us to harness the power of Intel hardware for accelerated computing.

To facilitate the learning process, we create workshops and learning materials that cover the topics outlined in the project proposal. These materials are developed using Jupyter Notebooks (.ipynb files), which can be executed locally on machines using Jupyter Notebook or Jupyter Lab. Furthermore, we ensure the accessibility and flexibility of our materials by making them compatible with cloud-based environments such as Google Colab.

In terms of frameworks and libraries, we utilize scikit-learn 1.2.2 for machine learning tasks, torch 2.0.0 for deep learning purposes, numpy 1.23.5 for numerical computations, pandas 2.0.2 for data manipulation and analysis, matplotlib 3.7.1 for data visualization, pydot 1.4.2 for graph visualization, graphviz 0.20.1 for generating graph structures, and keras 2.12.0 for building neural networks.

By employing these technologies and tools, we ensure that our project delivers a comprehensive learning experience, incorporating industry-standard frameworks and techniques. This approach enables students to gain practical knowledge and hands-on experience in utilizing cutting-edge technologies for AI-related tasks, equipping them with valuable skills for future endeavors in the field.

In our development methodology, we emphasize collaboration and teamwork to ensure the success of the project. We have assembled a diverse team with individuals possessing different expertise and backgrounds, contributing to a well-rounded approach to problem-solving.

The team members include:

• Murilo Boratto, who holds a Ph.D. in High-Performance Computing (HPC).
• Anúsio Menezes, who holds a Ph.D. in Artificial Intelligence (AI).
• Leonardo Rodrigues, a Bachelor of Engineering.
• Fernando Schettini, a student intern.
• Orlando Pires and Antônio Horácio, both students on scholarships.

To facilitate effective communication and progress tracking, we have scheduled regular status calls. These calls occur twice a week, on Mondays and Fridays, involving all team members: Murilo Boratto, Fernando Schettini, Anúsio, Antônio Horácio, and Orlando Pires. These calls serve as a platform for discussing project updates, sharing insights, and addressing any challenges encountered.

Furthermore, to ensure transparency and keep stakeholders informed, we have planned presentations of partial results at specific intervals. These presentations are scheduled for the months of June, August, and October. They provide an opportunity for the team to showcase their progress and receive feedback, ultimately enhancing the quality of the final deliverables.

By implementing this methodology, we foster a collaborative environment where knowledge and expertise are shared, allowing us to leverage the strengths of each team member. This approach promotes effective communication, timely feedback, and accountability, ultimately leading to the successful completion of the project.

Technologies Used

• python.
• Intel® Distribution for Python.
• Intel® oneAPI Toolkits.
• jupyter.
• scikit-learn 1.2.2.
• torch 2.0.0.
• numpy 1.23.5.
• pandas 2.0.2.
• matplotlib 3.7.1.
• pydot 1.4.2.
• graphviz 0.20.1.
• keras 2.12.0.

Documents and Presentations

Repository

https://github.com/muriloboratto/AI-intelOneAPI