AI is a once in a generation change in computing that’s expanding the capabilities of cloud server to the tiniest IoT device. Today, most ML is still performed on Arm CPUs and we continue to improve the efficiency of them, but dedicated Neural Network Processors -- NPUs -- are becoming more prevalent because they can dramatically improve the efficiency and performance of systems in a world where more and more ML must be run. For many years now, Arm has been on a mission to create the foundations to realize the opportunity of AI. The opportunities are massive, the market is open to everyone, and Arm is reducing risks by providing trusted blocks of technology, and through its ecosystem, delivering further tools and technologies to get to market faster and make it easier to unlock the value.

Arm Commercial and Marketing VP of ML Dennis Laudick will provide a comprehensive overview of Arm’s efforts and the opportunities created by AI and ML.

一般在訓練語音辨識系統時不僅需要蒐集大量的語音訊號，還需要人類對資料進行適當標註，此種方式限制了 AI 的使用範圍，以語音辨識為例，只有具備大量標註資料的中英文，能取得較高的辨識率。

在這個大數據的時代，若 AI 可以自行在網路上聆聽與閱讀大量未標註的語音及文字，將可改善蒐集資料所費不貲的問題。本演講將探討 AI 如何使用未標註的資料進行學習，並分享這些技術對 TinyML 所帶來的挑戰。

深入剖析目前主流智慧終端裝置的發展趨勢，以及 Arm 人工智慧解決方案如何適配多樣化的使用場景。演講中將實際展示說明 Arm 如何利用各式開源軟體開發 TinyML，並提升軟硬體效能以優化工作負載，協助生態系。

A combination of Machine Learning (ML) technology with microcontrollers (MCU) offers exciting opportunities for new applications. MCUs are typically memory and power constrained and offer a challenging environment for building ML applications. This presentation introduces the TFLite Micro (stands for TensorFlow Lite for Microcontrollers) as an open sourced framework that enables fast development of ML applications on microcontrollers. TFLite Micro runs on bare metal and is designed to be tiny in memory footprint. It is easy to modify and extend. Optimized CMSIS-NN kernels are available with TFLite Micro.

儘管人工智慧在伺服器上佔有優勢，但在物聯網裝置上仍有巨大的發展空間，對於物聯網開發人員來說，物聯網裝置的多樣性導致為 AI 應用找到合適的硬體和管理硬體資源既不容易亦不具效率。為此，Skymizer 開發了 Tiny ONNC，這是一款簡單高效的 AI 編譯器，由 Arm CMSIS-NN 驅動。Tiny ONNC 允許您使用您熟悉的工具，例如 Google Colab，用數據訓練模型，然後透過通用部署程序將模型部署到您的 IoT 裝置。

在本場演講中，您將了解：

語音辨識技術發展多年，近來隨著 AI 技術及雲端服務的發展而更廣受歡迎。除了雲端服務的方案，在許多資源有限的產品上同樣有強烈的本地語音辨識需求，然而目前開發人員並不容易找到可同時滿足辨識效能和成本考量的方案。有鑑於此，Cyberon 推出本地語音喚醒和指令辨識的演算法 DSpotter，成功移植到 ARMv8-M 架構的平台上，其需求低運算資源、高辨識率、高抗噪等特性，將能降低開發成本。

在本場演講中，您將了解：

As humans, we have the amazing ability to gain an understanding of the world around us by using our senses. By using our ears for example, we can quickly determine whether our car’s engine is having issues if we hear unusual engine noises, or by hearing someone coughing we can right away understand that someone is unwell. But humans can’t be everywhere and at all times. Can we enable machines to detect another machine’s failure simply by listening to it, sending preemptive alerts that minimize downtime, risk and production problems? Can it listen to footsteps in a restricted area?

Join us for a deep dive into how embedded machine learning can enable all Arm-powered devices to become smart.

During this session with practical presentation you will: