Silicon Labs AI/ML Development Solution
End-to-end machine learning (ML) development solution for existing Series 1 and Series 2 wireless SoCs and its new EFR32BG24 and EFR32MG24 products, which features integrated AI/ML hardware acceleration providing up to 4x faster interferencing with up to 6x lower power consumption for ML processing. The GSDK environment supports TensorFlow Lite-based flow to simplify the ML application development process. Using Silicon Labs’ ML development solution, designers can enhance embedded applications with AI/ML capabilities, even in ultra-low-power wireless IoT devices. ML computing at the edge enables a variety of smart industrial and home applications including sensor-data processing for anomaly detection, predictive maintenance, audio pattern recognition like glass-break sensing, simple-command word recognition, and vision use cases such as presence detection or people counting with low-resolution cameras.
AI at the Edge
Why Very Edge?
- Ever-increasing demand for small, integrated solutions
- High volume cost-sensitive markets require cost-effective edge solutions
- Battery powered devices need lower power consumption
- Small form factor requirements for size constraint devices
- Increase security: data never leaves the sensing device
Benefit Examples of Artificial intelligence and Machine Learning at the Very Edge
带宽得到优化
边缘设备传感器可生成大量原始数据，因而会占用大量带宽。Also, long-range, low-power communication could have limited bandwidth by default. AI/ML-enabled end nodes can pre-process data and transmit only what matters – helping to reduce bandwidth.
设计时间更快
Specialized AI modeling software create models that are used by small application MCUs, therefore avoiding complicated coding typically required to detect subtle differences in raw data.
设计尺寸更小与低功耗
AI/ML-based processing adds functional benefits and capabilities but without adding to the memory footprint or MCU requirements since code size tends to be reduced compared to traditional algorithms. Local processing also reduces current as radio communications are reduced.
Low Latency
Captured data is processed on the spot without sending to an aggregator on the network – this enables real-time operation.
Privacy, IP Protection & Security
Without sharing the vast majority of data outside of the device, bad actors have less data with which to engage in hacking activities. As raw data never leaves the device, Privacy and Intellectual Property protection is highly effective.
Offline Mode Operation
Since there is no need for external computing, local processing enables full offline-mode operation.
Cost Reduction
As data is being processed locally, the cost of data traffic, processing and storage could be significantly reduced.
应用示例
AI/ML Partners
什么是人工智能？
人工智能 (AI) 是试图模仿人类行为的系统，更具体而言，是对类似于人类行为的输入进行模拟响应的电气和/或机械实体。The best tangible example of this is voice recognition, where the system needs to understand colloquial terms, abbreviations, pronouns, as well as standard words to respond as if you were talking to your best friend. The crux of AI is taking input from something like a sensor or combination of sensors and determining an appropriate response based on a goal. 例如，家庭安全系统的目标是保护家庭。它必须确定来自振动和声音传感器的输入是否与破损窗户的输入相匹配，如果匹配，则触发警报并通知有关当局。它试图匹配假如您坐在沙发上并听到窗户破损时的行为——您会听到它、识别它并跑去拨打紧急服务电话。
什么是机器学习？
机器学习 (ML) 是系统在重复使用后自行改进的能力。其原理是，它可以使用正在收集的数据来自我改善提升。ML 是创建人工智能的副产品，因为研究人员需要一种方法来提升他们对输入的响应，即无需经常手动更新系统，而是让系统自行更新和改进。ML is typically a computer algorithm and is used to develop solutions like speech recognition.
人工神经网络
人工神经网络 (ANN) 是机器学习的一种实现方式，尽管它是具有多个层的非常高级的实现方式。Unlike ML, where it takes input and makes a decision in a single flow, an ANN has several nodes that each contribute to what to do based on the data. 更改一个节点的行为也会影响其他节点。这形成了与人类大脑非常相似的更复杂结构。
末端边缘
Starting at the bottom, the very end edge of AI/ML in the IoT consist of the simplest end device like small, low power sensors, simple smart home devices like lightbulbs, low-end thermostats and more. Typically, they use a microprocessor like a Cortex-M class and a very slimmed-down AI system to make the best decision based on the data they are sensing. 这可以使恒温器或门窗传感器了解它们所处的环境并做出最佳决策，而不是从工厂校准，因为如果它们部署在独特的环境中，则可能做出错误决策。这可以增强用户体验，让设备制造商和终端用户能够从设备中获取他们所需的一切。
中间边缘
在此级别上的 AI/ML 通常仍属于物联网网络的内部部署，但已转向诸如 Cortex-A 类的应用处理器。它们可以计算更多数据，并从所有其他终端节点或传感器获取输入。Here, you can make decisions based on the entire household or building instead of sending it to the cloud. 您可以让系统查看智能 HVAC 系统和智能照明系统的数据，以确保在用户离开建筑物的情况下灯会熄灭，并可能降低 HVAC 输出以节省电力。在这里，您首先开始桥接系统，在决定要做什么之前先收集所有输入数据。
顶部边缘
在顶部边缘，您现在要处理一些可能影响社区、城镇或城市的更具区域性的数据集和决策。You have even more powerful processors to handle the increase in data, which is typically municipalities and cities that are dealing with this layer to monitor resource use like electricity, parking space, and more.
云
The end all be all for data, it can seem like you have unlimited computational power, which isn’t far from the truth. 通常，即使决策不是在云级别做出，数据与决策至少会发送给云进行分析，以查看是否是正确的决策，如果不是，系统可以改进。This way, you can bridge many connected systems and manage entire fleets of assets. 这通常由大型实体使用，如零售店所有者、具有多阶段流程的工业参与者、连锁酒店等。
Development Tools (Coming Soon)
