티스토리 뷰

python & DS

IoT3

shannon. 2022. 7. 31. 19:37
반응형

Overview of external systems

Selecting transcript lines in this section will navigate to timestamp in the video
 
- [Narrator] External systems are one of the four main components of the IoT product. They include all the sources of data on the internet that can be accessed by an API. There are four main components to external systems. The first one is analytics, where data is transformed into information. The next is data services, these are sources of data available on the internet. Business systems provide internal operational data that could be useful to the product. And finally, the last source of data are other IoT products. It is this interfacing with these external sources of data that differentiates the IoT product from the connected product and the smart product, just like any other software on the internet. The virtualized IoT product can interface with these external systems the same way. This opens up massive amounts of functionality previously available only to software and in doing so, changes the definition of what a physical product is. All incremental value for an IoT product comes from transforming its data into useful information. External systems are one of the two main sources of data in IoT.
 
외부 시스템
 
외부 시스템은 IoT 제품의 4가지 주요 구성 요소 중 하나입니다.
여기에는 API로 액세스할 수 있는 인터넷의 모든 데이터 소스가 포함됩니다. 외부 시스템에는 네 가지 주요 구성 요소가 있습니다. 첫 번째는 데이터가 정보로 변환되는 분석입니다. 다음은 데이터 서비스입니다.
이들은 인터넷에서 사용할 수 있는 데이터 소스입니다. 비즈니스 시스템은 제품에 유용할 수 있는 내부 운영 데이터를 제공합니다. 그리고 마지막으로 데이터의 마지막 소스는 다른 IoT 제품입니다.
인터넷의 다른 소프트웨어와 마찬가지로 IoT 제품을 연결된 제품 및 스마트 제품과 차별화하는 것은 이러한 외부 데이터 소스와의 인터페이스입니다. 가상화된 IoT 제품은 이러한 외부 시스템과 동일한 방식으로 인터페이스할 수 있습니다. 이것은 이전에 소프트웨어에서만 사용할 수 있었던 엄청난 양의 기능을 열어주고 그렇게 함으로써 물리적 제품이 무엇인지에 대한 정의를 변경합니다. IoT 제품의 모든 증분 가치는 데이터를 유용한 정보로 변환하는 데서 나옵니다. 외부 시스템은 IoT의 두 가지 주요 데이터 소스 중 하나입니다.

Analytics and big data

Selecting transcript lines in this section will navigate to timestamp in the video
 
- [Instructor] Data analytics combines age-old statistical techniques with modern machine learning software. A classic example of statistics is linear regression, where a line is fit through a bunch of data points. But linear regression is also considered a supervised machine learning, or ML, algorithm. So is linear regression statistics or machine learning? It's both. If it's a static onetime event, it's statistics. If you're training a model with one dataset to be used to make predictions about another dataset and if the model is continuously being improved with new data and if the data sets are huge, it's machine learning. The trick in data analytics is to find the right model and then model its parameters, coefficients, and constants to improve the results. We can manage things if there are just two parameters, but if there's hundreds of parameters, which is often the case, or even hundreds of thousands of parameters, as found in deep learning networks, then we need machine learning to tweak the parameters, coefficients, and constants. Machine learning is simply statistics wrapped in learning software that tweaks the model's parameters to make it more accurate. Today's large datasets necessitate the tight integration of statistics and learning software. So although purists are technically correct when they say that not all analytics is machine learning, it's easy to understand why people use the terms interchangeably. Another way to classify analytics is by what it does. Analytics and ML identify things, like using facial recognition and speech recognition. They predict things, like predicting the effectiveness of a vaccine. They group things together, like when used in spam filtering, and they associate things, like when used for medical diagnoses. And yet another way to classify analytics, or in particular machine learning, is by how it learns. There's three types of learning software. There's supervised learning that requires the datasets to be labeled so the software can learn when it's right or wrong. This is used when identifying things and predicting things. A good example of this is teaching cars to identify things on the road using deep neural networks and learning the warning signs for a jet engine predictive maintenance system. A second learning software is unsupervised learning. Unsupervised learning only needs raw data. It is used for grouping things or associating things together, as it's used in recommender systems, for example, by Netflix and Amazon. The third way is reinforcement learning. Reinforcement learning needs the data and the rules. It's used for decision making as used in player AI for EA sports games and for the Roomba vacuum robot.
 
분석 및 빅데이터
데이터 분석은 오래된 통계 기술과 최신 기계 학습 소프트웨어를 결합합니다. 통계의 고전적인 예는 선형 회귀로, 선이 여러 데이터 포인트를 통해 적합합니다. 그러나 선형 회귀는 지도 머신 러닝(ML) 알고리즘으로도 간주됩니다. 선형 회귀 통계 또는 기계 학습입니까? 둘 다입니다. 정적 일회성 이벤트라면 통계입니다. 다른 데이터 세트에 대한 예측을 수행하는 데 사용할 하나의 데이터 세트로 모델을 훈련하고 새로운 데이터로 모델을 지속적으로 개선하고 데이터 세트가 방대하다면 머신 러닝입니다. 데이터 분석의 비결은 올바른 모델을 찾은 다음 해당 매개변수, 계수 및 상수를 모델링하여 결과를 개선하는 것입니다. 매개변수가 두 개뿐이면 모든 것을 관리할 수 있지만 수백 개의 매개변수(종종 있는 경우) 또는 수십만 개의 매개변수(딥 러닝 네트워크에서 볼 수 있음)가 있는 경우 매개변수, 계수를 조정하기 위해 기계 학습이 필요합니다. , 및 상수. 머신 러닝은 모델의 매개변수를 조정하여 더 정확하게 만드는 학습 소프트웨어에 싸인 통계입니다. 오늘날의 대규모 데이터 세트는 통계와 학습 소프트웨어의 긴밀한 통합을 필요로 합니다. 따라서 순수주의자들이 모든 분석이 기계 학습은 아니라고 말할 때 기술적으로 정확하지만 사람들이 이 용어를 같은 의미로 사용하는 이유를 이해하는 것은 쉽습니다. 분석을 분류하는 또 다른 방법은 수행하는 작업입니다. 분석 및 ML은 얼굴 인식 및 음성 인식 사용과 같은 항목을 식별합니다. 그들은 백신의 효과를 예측하는 것과 같은 것들을 예측합니다. 스팸 필터링에 사용되는 경우와 같이 항목을 그룹화하고 의료 진단에 사용되는 경우와 같은 항목을 연결합니다. 그리고 분석, 특히 기계 학습을 분류하는 또 다른 방법은 학습 방법입니다. 학습 소프트웨어에는 세 가지 유형이 있습니다. 데이터 세트에 레이블을 지정해야 하는 지도 학습이 있으므로 소프트웨어가 옳고 그름을 학습할 수 있습니다. 사물을 식별하고 사물을 예측할 때 사용합니다. 이에 대한 좋은 예는 심층 신경망을 사용하여 도로에서 사물을 식별하도록 자동차를 가르치고 제트 엔진 예측 유지 보수 시스템의 경고 신호를 학습하는 것입니다. 두 번째 학습 소프트웨어는 비지도 학습입니다. 비지도 학습에는 원시 데이터만 필요합니다. 예를 들어 Netflix 및 Amazon과 같은 추천 시스템에서 사용되는 것처럼 사물을 그룹화하거나 함께 연결하는 데 사용됩니다. 세 번째 방법은 강화 학습입니다. 강화 학습에는 데이터와 규칙이 필요합니다. EA 스포츠 게임용 플레이어 AI 및 룸바 진공 로봇에 사용되는 의사 결정에 사용됩니다.
 

Data services

Selecting transcript lines in this section will navigate to timestamp in the video
 
- [Narrator] Data Services, also known as data microservices, are abundant on the internet and until IoT, were only accessed by software. Data services such as weather, mapping and pricing can provide incredible value to physical products. Many data services are free, some are not. But what they have in common are, they're all accessed with an API. Once our product has been virtualized, it too can call APIs just like software. Consider IBMs Transformation from once a infrastructure company, to a software company, and then a professional service company and most recently with their acquisition of the weather company, and their increased emphasis on Watson, they've now become both a big data and analytics company. Weather isn't the only data IBM is analyzing. Watson is being used in consumer, commercial, industrial, and infrastructure IoT. The transformation of IBM clearly demonstrates where all companies should be heading. Since the IoT is virtualized in the software, it can interface with this rich form of data in the same way that internet software does today.
 
 

데이터 서비스 이 섹션에서 스크립트 라인을 선택하면 비디오의 타임스탬프로 이동합니다. - [내레이터] 데이터 마이크로서비스라고도 알려진 데이터 서비스는 인터넷에 풍부하게 존재하며 IoT까지는 소프트웨어로만 액세스할 수 있었습니다. 날씨, 매핑 및 가격 책정과 같은 데이터 서비스는 실제 제품에 놀라운 가치를 제공할 수 있습니다. 많은 데이터 서비스가 무료이지만 일부는 그렇지 않습니다. 그러나 공통점은 모두 API로 액세스할 수 있다는 것입니다. 우리 제품이 가상화되면 소프트웨어처럼 API를 호출할 수 있습니다. 한때 인프라 회사에서 소프트웨어 회사, 그 다음에는 전문 서비스 회사로, 그리고 가장 최근에는 기상 회사를 인수하고 Watson에 대한 강조를 강화하면서 이제 빅 데이터 및 분석 회사가 된 IBM의 혁신을 고려하십시오. . 날씨는 IBM이 분석하는 유일한 데이터가 아닙니다. Watson은 소비자, 상업, 산업 및 인프라 IoT에서 사용되고 있습니다. IBM의 변화는 모든 기업이 나아가야 할 방향을 명확하게 보여줍니다. IoT는 소프트웨어에서 가상화되기 때문에 오늘날 인터넷 소프트웨어와 동일한 방식으로 이 풍부한 데이터 형식과 인터페이스할 수 있습니다.

 

 

Business systems

Selecting transcript lines in this section will navigate to timestamp in the video
 
- The Internet of Things is changing business processes through the automated collection of business intelligence. Now IoT products can interface directly with business systems, circumventing the need to manually input and extract data. Interfacing products directly with the CRM system or Customer Relationship Management System enables the enterprise to track customer requirements and needs. If, for example, a tire manufacturer senses tire tread, they can notify the fleet manager when a tire is becoming unsafe and offer them a deal to replace it. Interfacing products directly with a PLM system or Product Lifecycle Management system can track product use and use this information to prioritize future features. If, for example, a lot of customers use their Smartwatch in the water, the manufacturer may prioritize the development of a water sports version. Interfacing products directly with the Supply Chain Management system, they can follow their progress through the entire supply chain. For example, since Tesla cars can be ordered online, the company can track each car from the time of order through manufacturing to delivery. Interfacing products directly with an ERP system or Enterprise Resource Planning System enables enterprises to track workforce efficiency. For example, a smart factory can track how long each worker uses each machine. Through analytics, the enterprise can determine its most efficient staffing. This direct interfacing with business systems through their APIs is, in itself, a boost in operational efficiency. But it also provides IoT products with rich sources of internal business data unavailable to traditional products.
 
비즈니스 시스템 이 섹션에서 스크립트 라인을 선택하면 비디오의 타임스탬프로 이동합니다. - 사물 인터넷은 비즈니스 인텔리전스의 자동화된 수집을 통해 비즈니스 프로세스를 변화시키고 있습니다. 이제 IoT 제품은 비즈니스 시스템과 직접 인터페이스하여 데이터를 수동으로 입력하고 추출할 필요가 없습니다. 제품을 CRM 시스템 또는 고객 관계 관리 시스템과 직접 연결하면 기업이 고객 요구 사항과 요구 사항을 추적할 수 있습니다. 예를 들어 타이어 제조업체가 타이어 트레드를 감지하면 타이어가 안전하지 않게 될 때 차량 관리자에게 알리고 교체 계약을 제안할 수 있습니다. 제품을 PLM 시스템 또는 제품 수명 주기 관리 시스템과 직접 연결하면 제품 사용을 추적하고 이 정보를 사용하여 향후 기능의 우선 순위를 지정할 수 있습니다. 예를 들어 많은 고객이 물 속에서 Smartwatch를 사용하는 경우 제조업체는 수상 스포츠 버전의 개발을 우선시할 수 있습니다. 제품을 공급망 관리 시스템과 직접 인터페이스하여 전체 공급망에서 진행 상황을 추적할 수 있습니다. 예를 들어 Tesla 자동차는 온라인으로 주문할 수 있기 때문에 회사는 주문 시점부터 제조, 배송까지 각 자동차를 추적할 수 있습니다. ERP 시스템 또는 전사적 자원 관리 시스템과 직접 제품을 연결하면 기업이 인력 효율성을 추적할 수 있습니다. 예를 들어, 스마트 공장은 각 작업자가 각 기계를 사용하는 시간을 추적할 수 있습니다. 기업은 분석을 통해 가장 효율적인 인력 배치를 결정할 수 있습니다. API를 통한 비즈니스 시스템과의 직접 인터페이스는 그 자체로 운영 효율성을 향상시킵니다. 그러나 또한 기존 제품에서는 사용할 수 없는 풍부한 내부 비즈니스 데이터 소스를 IoT 제품에 제공합니다.
 

Other IoT products

Selecting transcript lines in this section will navigate to timestamp in the video
 
- [Instructor] IoT products are external systems. Each generates data, which theoretically means they can be connected to each other like Lego blocks to deliver outcomes. The IoT platform is the technical underpinning that allows one product to be connected to another. Over time, that product will develop into a family of interconnected products. When companies can orchestrate their products to work together, they can deliver their customers' outcomes. One way to quantify the value of multiple connected products is with Metcalfe's law. Metcalfe's law states the value of a network increases exponentially with the number of its sources. Since each product is a data source, value goes up exponentially in proportion to the size of the product network. This together leaves us to the outcome economy. Like with Legos, the more parts to the system, the more value there is.
 
기타 IoT 제품 
 
IoT 제품은 외부 시스템입니다. 각각은 데이터를 생성하므로 이론적으로 레고 블록처럼 서로 연결하여 결과를 제공할 수 있습니다. IoT 플랫폼은 하나의 제품을 다른 제품과 연결할 수 있도록 하는 기술적 토대입니다. 시간이 지남에 따라 해당 제품은 상호 연결된 제품군으로 발전할 것입니다. 회사가 제품을 함께 작동하도록 오케스트레이션할 수 있을 때 고객의 결과를 제공할 수 있습니다. 연결된 여러 제품의 가치를 수량화하는 한 가지 방법은 Metcalfe의 법칙을 사용하는 것입니다. Metcalfe의 법칙에 따르면 네트워크의 가치는 소스의 수에 따라 기하급수적으로 증가합니다. 각 제품은 데이터 소스이기 때문에 가치는 제품 네트워크의 크기에 비례하여 기하급수적으로 증가합니다. 이것은 함께 우리를 결과 경제로 남깁니다. 레고와 마찬가지로 시스템에 더 많은 부품이 포함될수록 더 많은 가치가 있습니다.
반응형

'python & DS' 카테고리의 다른 글

dbscan  (0) 2023.08.24
missing value 처리 방식 실제사례  (0) 2023.08.14
IoT2  (0) 2022.07.31
IoT  (0) 2022.07.31
conclusion  (0) 2022.07.31
공지사항
최근에 올라온 글
최근에 달린 댓글
Total
Today
Yesterday
링크
«   2024/07   »
1 2 3 4 5 6
7 8 9 10 11 12 13
14 15 16 17 18 19 20
21 22 23 24 25 26 27
28 29 30 31
글 보관함