Day 1 & Day 2 Time Table (Sep. 12 & 13)

NOTE: For presenters, please either have Lark installed on your laptop to share your screen during your presentation OR send us your slides to use the shared laptop during the event.

You need to login with your LFID to edit the confluence page. If you don't have one, you can create one here:

https://openprofile.dev/

Day1 (Sep. 12)

Lark Meeting: https://vc-us.larkoffice.com/j/671149719

YouTube Live: https://www.youtube.com/watch?v=HEWWkO7GLM8

Summary

The meeting discussed AI and edge computing, the main contents included:

• AI applications: in various fields like Auto SE Kim, WASM runtime, and China Mobile's strategic planning.

• Challenges and opportunities: in model coupling, edge computing, and industry collaboration.

• Demonstrations: showcasing the capabilities of these technologies.

• Open source projects and tools: such as ByteDance's Babid Multimedia Framework and BMF libraries integrated with Hugging Face Library.

• AI networking innovations: and the role of telcos in hosting AI services.

• Upcoming event: "futures in Taipei".

Chapters

00:00 Advances and Applications of AI in Edge Computing and Distributed Cloud

This section features presentations on AI agent building and capabilities. Gary from codes discusses its various uses and how to build one. Tina Tsou elaborates on bridging the gap between edge AI and distributed cloud computing, covering challenges, tools, security, and future prospects. There's also mention of related workstreams, projects, and opportunities for collaboration.

38:21 Edge and Cloud Collaborative Support and Security for Large Language Models

This section focuses on the deployment and optimization of edge and cloud inferences for large language models. It covers various deployment models, model conversion and optimization, reduction of model size for edge devices, scheduling for fast execution, and a framework for attestation and evidence support. Different techniques and challenges are discussed along with solutions.

01:20:19 The Future of Network and Computing Infrastructure: Insights and Initiatives by Harohisa Fukano from Fujitsu

This section is about Harohisa Fukano from Fujitsu presenting on the future of network and computing infrastructure for the 2030s. He covers LFH projects, computing challenges like power consumption and flexibility, the bottleneck disaggregated computer, joint efforts with the Ion Global Forum, and potential use cases including generative AI plus robotics and green energy data processing, along with upcoming POC phases.

01:50:24 Edge Computing, IoT, and Function Calling in AI Projects Presentation and Demos

This section presents edge computing, IoT, and function calling. Tom introduces Shifu for IoT, detailing its architecture, use cases like with the California Strawberry Commission and a supermarket chain. CC discusses Geo distributed computing, function calling, and related improvements. Demos of Shifu integration and function calling are shown. Questions and discussions follow.

02:30:31 Discussions on Various Topics including Technology, Business, and Education

This section is a complex and lengthy discussion covering various topics including lunch and networking plans, business models, technology applications like TikTok, issues related to code and software development, legal aspects in certain fields, and experiences with different systems and services. The conversation also touches on teams, support, and various challenges and opportunities.

03:32:20 Overview of Auto SE and Comparisons in AI Application Workflow

This section mainly focuses on various discussions related to Auto SE and its features. Li Chuan presents on Auto SE, including its capabilities, comparison with Github Pilot, key highlights like link and live features, switch setup actions, direct support, standing in SMB benchmark leaderboard, and cost savings compared to SMB agent. Also, there are mentions of future discussions and ongoing targeting.

04:14:53 The Need for Tightly Coupled Language Models and Web Assembly in Applications

This section is mainly presented by Michael Yu Yan. He discusses the need for tightly coupling large language models with specific knowledge bases and applications. He gives examples like the chemistry and Rust programming demos, highlighting issues with Python and cross-platform compatibility. He also mentions web assembly as a solution and showcases a cross-GPU demo. Everything presented is open source.

04:56:05 Discussions on AI Applications and Open Source Projects in Various Contexts

This section includes various discussions on AI-related topics. It covers China Mobile's strategic planning in the AI era, such as AI in operation services, creativity, new quality productivity, and strategic decision making. It also features ByteDance's open source projects and their potential in addressing AI challenges. Sujata Tibrewala from ByteDance discussed AI's impact on jobs and the environment.

06:38:08 Advancement and Comparison of AI Networking Technologies and Protocols

This section discusses advancements in AI networking, focusing on scale up and scale out networking. It explores protocols like TCPIP and RDMA, links such as NV Link and CXL, and the need for efficient data transfer in GPU clusters. The presenter also compares different technologies, emphasizing low latency and high performance, and considers future positions regarding networking architectures and protocols.

07:12:51 Overview and Use Cases of Ion Global Forum and Upcoming Event

This section involves Rintaro Harada from NTT presenting about the Ion Global Forum. He introduces its overview, including the aim to create a sustainable society, its founding by NTT, Intel, and Sony. Discusses goals, technical aspects, POC activities, use cases like Smart City, and potential applications in various industries. Also announces an upcoming event in Taipei next month.

Day2 (Sep. 13)

Lark Meeting: https://www.us.larkoffice.com/calendar/share?token=e8edd94e9827b518b68b279ca5120824

Youtube Live: https://www.youtube.com/watch?v=c-PUUrwWeK0

Summary

The meeting discussed the application of AI in various fields, the main contents included:

• Speech recognition: Wilson Wang introduced the challenges and solutions in speech recognition, including the use of proximity addressable memory for efficient processing.

• Data catalog and LM agent: Hao Xu presented the data catalog and LM agent, which can help users query data and interact with LM.

• Conversational assistance interoperability: Deborah Dahl talked about enabling voice and conversational AI systems to function like the web, and the Open Voice Interoperability Initiative.

• Sound alike errors: Wilson Wang focused on the problem of sound alike errors in speech recognition and how to fix them.

• Intermediate stage in SLM: Wilson Wang emphasized the need for an intermediate stage before the SLM to identify likely possible sound like substitutions.

Chapters

00:00 Meeting Kickoff

Meeting Kickoff

01:34 Challenges and Solutions in Real-Time Speech Recognition with AI and Energy Efficiency Considerations

This section discusses speech recognition challenges, especially sound alike errors and precision issues. It focuses on post-processing with AI, using small language models and proximity addressable memory for accuracy. It also covers hardware requirements, energy efficiency concerns, and the status of ongoing efforts and potential future developments.

24:14 Challenges and Approaches in Context Processing and Security for First Responders' Devices

This section discusses challenges and approaches in language processing for various scenarios like factories and first responders. It considers the need for context base processing, testing in different conditions, issues of security and privacy, authentication, and the use of models like Lama 2. The conversation also touches on potential solutions and the next steps for testing and implementation.

32:37 The Project of Using LM Agent to Address Data Query Issues in the Team

This section is about a presentation by Hao Xu representing the item team. They have around 4.5 people working on the project. The team is addressing the issue of numerous ad hoc data-related queries and exploring using an LM agent to assist. They avoid giving raw data to third-party vendors due to privacy concerns and are building a system with three major components for processing and answering queries.

40:52 Overview of Components and Features in Data Query and Processing Framework

This section covers various aspects including direct data query methods, use of data frame APIs and Linux projects for metadata. It discusses the core component agent controlling aspects like prompts and tools. Workflow customization, built-in tools, prompt templates, vector stores, data catalogs, and their relationships are also mentioned.

49:49 Overview of Hao Xu's Project and Call for Contributions

This section covers the use of Python and various technologies for building an application. It mentions an integrated notebook environment for query modification and interaction. Three teams are currently using ITA. The project is in an early stage, lacks cloud deployment, needs robust testing and automation. It was recently renamed and those interested can reach out. There are no questions today and a next person introduction is planned.

56:32 Interoperability of Conversational AI Systems and Ongoing Work

This section is about Deborah Dahl from Conversational Technologies discussing the Open Voice Interoperability Initiative. She explains the aim to enable collaboration among conversational AIs, reducing complexity and duplication. Different use cases are mentioned, and the focus is on developing interoperability standards. Messages and capabilities for chatbots are detailed, along with ongoing work like enabling teams, providing context, and ensuring data security. She invites feedback and participation.