Position: 吳俊逸 > AI
AI in industry
by 吳俊逸 2021-09-11 11:59:38, Reply(0), Views(1312)

Samsung SDS Brightics, an AI Accelerator for Automating and Accelerating Deep Learning Training

Training AI models is an extremely time-consuming process. Without proper insight into a feasible alternative to time-consuming development and migration of model training to exploit the power of large, distributed clusters, training projects remain considerably long lasting. To address these issues, Samsung SDS developed the Brightics AI Accelerator. The Kubernetes-based, containerized application, is now available on the NVIDIA NGC catalog – a GPU-optimized hub for AI and HPC containers, pre-trained models, industry SDKs, and Helm charts that helps simplify and accelerate AI development and deployment processes.  

The Samsung SDS Brightics AI Accelerator application automates machine learning, speeds up model training and improves model accuracy with key features such as automated feature engineering, model selection, and hyper-parameter tuning without requiring infrastructure development and deployment expertise. Brightics AI Accelerator can be used in many industries such as healthcare, manufacturing, retail, automotive and across different use cases spanning computer vision, natural language processing and more.  

Key Features and Benefits: 

  • Is case agnostic and covers training all AI models by applying autoML to tabular, CSV, time-series, image or natural language data to enable analytics; image classification, detection, and segmentation; and NLP use cases. 
  • Offers model portability between cloud and on-prem data centers and provides a unified interface for orchestrating large, distributed clusters to train deep learning models using Tensorflow, Keras and PyTorch frameworks as well as autoML using SciKit-Learn.
  • AutoML software automates and accelerates model training on tabular data by using automated model selection from Scikit-Learn, automated feature synthesis, and hyper-parameter search optimization. 
  • Automated Deep Learning (AutoDL) software automates and accelerates deep learning model training using data-parallel, distributed synchronous Horovod Ring-All-Reduce Keras, TensorFlow, and PyTorch frameworks with minimal code. AutoDL exploits up to 512 NVIDIA GPUs per training job to produce a model in 1 hour versus 3 weeks using traditional methods. 
Figure 1. The amount of time it takes to train 1 iteration of a ResNet 50 image classification model using a normal 8 GPU machine is about 504 hours. Using the enhanced inter-GPU communication of Brightics AI Accelerator lowers this to about 4 hours using 128 GPUs. In total, this represents a speed improvement of 126 times.

Integrated energy management system based on data analytics

“Since real-time monitoring for energy consumption status is possible, we can easily find out waste factors of every production process & facility and promptly response."


This electronics manufacturer pays USD 135 million for year as an energy expense. Also, energy efficiency was an important task to secure competitiveness in manufacturing as their energy cost kept increasing. 

The partner thought it is essential to do the activities to analyze the energy consumption status and energy efficiency by process and facility. 
However, the partner is producing various products and has a variety of facilities in a large workplace. 
That disables the partner to respond to the challenges with limited on-site activities by relevant workforce.  

To that end, there was a need for the system to analyze the energy consumption status and energy efficiency real time. 
The partner also wanted to conduct sustainable activities for energy reduction by analyzing causes of efficiency change and establishing an immediate response system. 


Energy-related data integration and integrated energy management system for the entire company have been implemented.  

[Energy monitoring] 
ㆍ Automatically gather for energy usage & operation data, enhance data integrity 
ㆍ Visualize real-time energy usage by process and facility 
ㆍ Visualize energy intensity by business division, factory and product 

[Analyze energy efficiency & cause of fluctuation] 
ㆍ Develop models to analyze facilities that consume a lot of excessive energy 
ㆍ Draw efficiency variables and optimal values 

[Establish management systems for energy estimation] 
ㆍ Predict energy consumption based on energy usage data analytics 
ㆍ Establish energy reduction plan & activities based on forecast value 


SDS Energy Service 
Samsung SDS provides services from energy consulting to total services regarding infrastructure and system development depending on business type. 

■ Energy consulting: energy diagnosis and analysis of consumption patterns, establishment of optimization measures 
■ Measuring infrastructure establishment: establishment infrastructure as well as measures for IoT-based infrastructure to effectively manage energy 
■ Integrated data management : Data gathering, cleansing and saving from various IoT equipment/systems 
■ Data analysis: Data modeling and predictive simulation for AI-based efficiency analysis 
■ Integrated management system implementation: development of control systems for energy plan/performance management, energy efficiency analysis and optimal operation 


Partners have secured the ground for energy reduction through energy visualization at workplaces, which enables them to have reasonable operation of the facilities and systems. 


DeepMind AI Reduces Google Data Centre Cooling Bill by 40%

Reducing energy usage has been a major focus for us over the past  10 years: we have built our own super-efficient servers at Google, invented more efficient ways to cool our data centres and invested heavily in green energy sources, with the goal of being powered 100 percent by renewable energy. Compared to five years ago, we now get around 3.5 times the computing power out of the same amount of energy, and we continue to make many improvements each year.

Major breakthroughs, however, are few and far between - which is why we are excited to share that by applying DeepMind’s machine learning to our own Google data centres, we’ve managed to reduce the amount of energy we use for cooling by up to 40 percent. In any large scale energy-consuming environment, this would be a huge improvement. Given how sophisticated Google’s data centres are already, it’s a phenomenal step forward.

The implications are significant for Google’s data centres, given its potential to greatly improve energy efficiency and reduce emissions overall. This will also help other companies who run on Google’s cloud to improve their own energy efficiency. While Google is only one of many data centre operators in the world, many are not powered by renewable energy as we are. Every improvement in data centre efficiency reduces total emissions into our environment and with technology like DeepMind’s, we can use machine learning to consume less energy and help address one of the biggest challenges of all - climate change.

One of the primary sources of energy use in the data centre environment is cooling. Just as your laptop generates a lot of heat, our data centres - which contain servers powering Google Search, Gmail, YouTube, etc. - also generate a lot of heat that must be removed to keep the servers running. This cooling is typically accomplished via large industrial equipment such as pumps, chillers and cooling towers. However, dynamic environments like data centres make it difficult to operate optimally for several reasons:

  1. The equipment, how we operate that equipment, and the environment interact with each other in complex, nonlinear ways. Traditional formula-based engineering and human intuition often do not capture these interactions.
  2. The system cannot adapt quickly to internal or external changes (like the weather). This is because we cannot come up with rules and heuristics for every operating scenario.
  3. Each data centre has a unique architecture and environment. A custom-tuned model for one system may not be applicable to another. Therefore, a general intelligence framework is needed to understand the data centre’s interactions.

To address this problem, we began applying machine learning two years ago to operate our data centres more efficiently. And over the past few months, DeepMind researchers began working with Google’s data centre team to significantly improve the system’s utility. Using a system of neural networks trained on different operating scenarios and parameters within our data centres, we created a more efficient and adaptive framework to understand data centre dynamics and optimize efficiency.

We accomplished this by taking the historical data that had already been collected by thousands of sensors within the data centre - data such as temperatures, power, pump speeds, setpoints, etc. - and using it to train an ensemble of deep neural networks. Since our objective was to improve data centre energy efficiency, we trained the neural networks on the average future PUE (Power Usage Effectiveness), which is defined as the ratio of the total building energy usage to the IT energy usage. We then trained two additional ensembles of deep neural networks to predict the future temperature and pressure of the data centre over the next hour. The purpose of these predictions is to simulate the recommended actions from the PUE model, to ensure that we do not go beyond any operating constraints.

We tested our model by deploying on a live data centre. The graph below shows a typical day of testing, including when we turned the machine learning recommendations on, and when we turned them off.

Our machine learning system was able to consistently achieve a 40 percent reduction in the amount of energy used for cooling, which equates to a 15 percent reduction in overall PUE overhead after accounting for electrical losses and other non-cooling inefficiencies. It also produced the lowest PUE the site had ever seen.

Because the algorithm is a general-purpose framework to understand complex dynamics, we plan to apply this to other challenges in the data centre environment and beyond in the coming months. Possible applications of this technology include improving power plant conversion efficiency (getting more energy from the same unit of input), reducing semiconductor manufacturing energy and water usage, or helping manufacturing facilities increase throughput.


How AI can be used in HR

Hire: Efficient and effective recruitment 
The job of a recruiter is time pressured and complex, often having to fill many roles at once. Recruiters need to prioritize all of the different roles they are responsible for, and at the same time, they need a way to differentiate among candidates competing for the same role. Not meeting these challenges effectively enough can mean the wrong roles get prioritized, and even where the right roles are prioritized, the wrong candidates might be selected for roles. AI can be used in this setting to predict how long a job requisition will take to fill based on historical data, allowing recruiters to reprioritize as needed. AI can also be used to determine the match between a candidate’s resume and the job requisition, and to make accurate predictions of future performance based on information about the candidate collected in the job application process. Furthermore, it can help recruiters write more inclusive job descriptions and filter candidates more effectively, minimizing the impact of unconscious bias in their process and practices.

Deploying AI in recruitment allows faster and more accurate hiring, and a better candidate and recruiter experience.

AI recruitment at IBM 
In a large organization like IBM, effective prioritization of recruitment demands careful selection of applicants. IBM needed a better way to help recruiters surface the top candidates for open jobs and to prioritize the most important requisitions. The solution developed, IBM Watson Recruitment (IWR), uses AI to leverage information about the job market and past experiences of hiring candidates to predict time to fill and identify the candidates most likely to be successful. By helping the recruiter prioritize and rank candidate suitability, AI frees up time to focus on the core of recruiting: building and nurturing relationships with candidates. AI derives required skills from job requisitions and generates a match score against skills described in resumes. The solution can also generate a predictive score based on biographical data (e.g., whether or not they have led a team) in the resume. These scores predict future job performance. Importantly, IWR monitors hiring decisions to make sure they are free from bias. In summary, deploying AI in the recruitment function allows faster and more accurate hiring, and a better candidate and recruiter experience.

AI-supported compensation planning at IBM 
Making complex compensation decisions accurately across an organization is a challenge, and one that IBM uses AI to address. IBM designed an AI-powered decision support tool that assists with compensation planning, helping managers avoid underweighting or overweighting the critical data points. The application reviews dozens of data points in making its recommendations, integrating external information from sources like the Bureau of Labor Statistics with internal data on factors such as cost to replace. The application is currently being deployed for tens of thousands of first-line managers to assist with their compensation planning, following successful early trials in focused geographies. Importantly, when using the tool, managers have the opportunity to override the AI recommendation about any given employee, and the system can continue to learn from managers’ actual decisions. In general, managers tend to follow the recommendations the AI provides, and this has helped ensure employees are not overpaid or underpaid at IBM. IBM also emphasizes transparency in AI-based compensation support: employees can see where they sit relative to the market, because the low and high range of compensation for workers with their skills is provided, in addition to their personal salary.

AI chatbot use in IBM HR 
IBM has chatbots that are used by employees all year round in areas across HR. For example, HR has deployed chatbots to support employees with their benefits enrollment decisions, and to support managers with their compensation planning; both are areas with designated time periods characterized by high usage rates, requiring fast responses to user questions. Chatbots that are busier at certain times of the year, such as the performance management, benefits enrollment, and compensation planning chatbots, are considered ‘seasonal bots.’ IBM also has bots that are accessed 24 hours a day, seven days a week, year-round. An example is IBM’s popular new-hire chatbot. It is one of the busiest chatbots at IBM, answering 700 questions a day. New hire chatbots are particularly helpful because they resolve the challenge of not knowing who to ask for help. IBM’s goal with chatbots is to get answers to employees quickly and accurately, while reducing the amount of effort it takes to support HR programs. The time saved can then be spent on experts answering more complex questions and problems about HR issues.

IBM artificial intelligence can predict with 95% accuracy which workers are about to quit their jobs

IBM receives more than 8,000 resumes a day, IBM HR has a patent for its "predictive attrition program" which was developed with Watson to predict employee flight risk and prescribe actions for managers to engage employees.A clearer career path is needed for many. Among the tasks that HR departments and corporate managers have not always proved effective at, and where AI will play a bigger role in the future, is keeping employees on a clear career path and identifying their skills. IBM technology can view the tasks employees are completing, the educational courses they have taken and any rankings they have earned. Through these data points, the AI skills inference and HR managers can gain a greater understanding of an employee's skill set than they would by assessing the feedback from manager surveys.