Reducing Workload by Defective Product Classification

Total Page:16

File Type:pdf, Size:1020Kb

Reducing Workload by Defective Product Classification Reducing Workload by Defective Product Classification Author Date Published Dr. Nguyen Ngoc Tam February 7th, 2020 ContentAbstract 3 Introduction 4 Background 5 Solution 7 Results 10 Conclusion 11 How can FPT help? 13 About the author 13 Reducing Workload by Defective Product Classification 2 AbstractPlaying the roles of eyes and ears in the manufacturing process, Artificial Intelligence/Machine Learning (AI/ML) is contributing to addressing the growing challenges of product quality control. According to the corporate insurance carrier, AGCS, “defective products not only pose a serious safety risk to the public but can also cause significant financial and reputational damage to the companies concerned. Defective product incidents have caused insured losses in excess of $2B over the past five years, making them the largest generator of liability losses.” With over 50% of manufacturers planning to increase Artificial Intelligence/ Machine Learning spending in the coming years, according to Forbes Insights research, the industry is leaving behind its stagnant reputation to dive into automation. Reducing Workload by Defective Product Classification 3 IntroductionA very common challenge of ML application in manufacturing is data acquisition and data processing. While the accuracy of data plays a strong role in the performance of ML algorithms, the challenge is to go through massive data sets and label each unit with a high level of accuracy. This paper sought to solve the problem of using AI/ML in the process of defective product classification by using Autoencoder model and One-class classification algorithm to classify normal and abnormal ones. The solution does not only help to extract important features, thereby precisely detecting abnormalities, but also reduces the amount of time spent on these tasks. Experiments on client’s data set showed a reduction of approximately 35% on user workload, allowing manufacturers to achieve higher productivity and lower burden on quality control workers. Reducing Workload by Defective Product Classification 4 BackgroundProduct quality control plays an important role in maintaining and strengthening the link between manufacturers and final customers. Statistics show that 45% of companies using smart manufacturing technologies have experienced increased customer satisfaction. The visual quality impression of a product has a strong influence on the decision of whether a product is purchased or not. Traditionally, the verification process, when manually handled by human eyes, is time-consuming and takes Usually, a lot of effort. Today, the automated verification system using artificial intelligence is becoming more and more defects occur popular among smart technology manufacturers. very rarely, A common approach using artificial intelligence for the verification process is to use supervised learning so experts model. However, the challenge of this approach is need to spend in the training process of faulty sample collection. Usually, defects occur very rarely, so experts need extreme labor to spend extreme labor effort to label sample into a unit level. To overcome this challenge, this paper effort to label proposes an unsupervised learning approach for sample into a defect detection, which can save much more human labor effort during the training process. unit level. This unsupervised learning model uses Autoencoder network, which will reconstruct the normal input image based on a sample of normal data, and if it meets the abnormal data, it will not reconstruct defect areas. This process will leave images after the reconstruction process with strange bright spots on the CT image. By subtracting the corresponding query region, a pixel-wise anomaly score is obtained, which is then used to detect defects, therefore experts can save time going through all the normal sample. This approach however, still has some limitations: In some cases, anomalies are small bright dots which hardly recognizable even by human eyes. There is also no apparent difference between the abnormal image and the normal image. Patterns of data set vary, which causes difficulty for the Autoencoder network to learn the main patterns. To overcome the above limitations, this paper proposes a combination model in which the outcome of Autoencoder network will be the input of One-class model to identify objects in a specific concept. This combination approach will be able to reduce the number of images that need to be inspected hence save more human workload and make it easier for experts to detect defect sample. Reducing Workload by Defective Product Classification 5 This combination approach will be able to reduce the number of images that need to be inspected, and hence save more human workload as well as make it easier for experts to detect defect sample. Reducing Workload by Defective Product Classification 6 Solution The solution is the combination of an Autoencoder network and a One-class model in training models. The idea behind this approach is like how we break an object into multiple pieces and then reconstruct the object in a predefined way, if the object is not the same then it is defected. Fig 1 shows the overall architecture of this model in the training and testing phases, which includes: Training Phase This phase will help AI/ML to learn how to classify normal and abnormal data. Normal process in supervised model would require manual effort to label the data to help improve the model. With the unsupervised approach, U-net will improve the model overtime. Testing Phase In this phase, the learned classifiers① model from training phase, which are the output of the One-class classification with Local outlier factor algorithm, will be used to discriminate between normal images and abnormal images. Overtime when the model is improved through the 2 phases, the accuracy output of image processing will also improve. In this paper, we run the testing phase on the real data set of 25,088 images from our client. ②Reducing Workload by Defective Product Classification 7 TRAINING PROCESS Loss SSIM U-net Training AutoEncoder Encode Decode Same batch Shared data weights Training One Class Encode Encoder Model 1×256 Compactness Loss Hyper Sphere (boundary) TESTING PROCESS Hyper Sphere (boundary) Normal Images Unet - Encoder Normal One Class Encode Model Abnormal Abnormal Images Figure 1: Overall architecture of the proposed model The training is conducted separately by using U-net and One-class model. We illustrate the procedures as follows: Step 1 Step 2 Step 3 For each batch sample, Encode block was trained Use the Encode block training U-net with SSIM with compactness loss. that was trained in Step loss to update weight Training Autoencoder 1 as a pre-train model for for the Autoencoder. with SSIM loss function training the One-class SSIM loss function plays and encoder with model to identify the the role to improve the Compactness loss is normal objects out of prediction accuracy of carried out in parallel. the data set. the model. The lower the This step aims to drag weight loss, the better the characteristics of the accuracy the model will image in the same class achieve. as close as possible. Reducing Workload by Defective Product Classification 8 An autoencoder is a neural network that consists of Autoencoders 3 main parts: The encoder, the latent representation with U-net and the decoder. It learns to reconstruct an input image x⋹Rk×h×w to its output. Therefore, Autoencoders are unsupervised learning models (do not require labelled inputs to enable learning). An Autoencoder consists of an encoder function E: Rk×h×w → Rd and a decoder function D: Rd → Rk×h×w, where d denotes the dimensionality of the latent space and k,h,w denote the number of channels, height, and width of the input image, respectively. In our experiments, the functions E and D are parameterized by U-net and SSIM loss function. The other evaluation metrics are all based on per- pixel that assume unrealistic independence between SSIM loss neighbouring pixels. Therefore, they fail to detect function structural differences between the inputs and their reconstructions. By adapting the loss and evaluation functions to capture local inter-dependencies between image, we specifically motivate the use of the structural similarity metric SSIM (x,y) as both the loss and evaluation metric for Autoencoder. To evaluate local dependence between image regions, SSIM (x,y) is used as a loss and evaluation metric for Autoencoder. All data used in training are the vector outputs of Compactness loss the encoder network which belong to the same class. Therefore, they all share the same class label. Compactness loss calculates the average similarity between the constituent samples of a given batch. We use this loss function to drag the characteristics of the image in the same class as close as possible. Reducing Workload by Defective Product Classification 9 Results We define the most important class which must predict exactly is Positive (P-anomaly), and Negative (N-normal). Values that are correctly classified by the model are labelled true (T), while incorrect predictions are false (F). Our focus is to make sure that an abnormal image was not predicted as normal. Therefore, we define the ratio betweenTN (True Negative) and the total normal images as the user workload reduction. Instead of checking all the images, our method proposes people to only check images which are considered abnormal by model. TN workload reduction = total of normal images testing During the testing phase, a data set includes 25088 normal images and 92 abnormal images have been used. The model predicts exactly 8724 normal images out of total 25088 normal images, leaving 16,364 normal images that require manual verification. The manual workload reduction in the above test is approximated 35%: 8724 workload reduction = = 3 25088 In a larger scale of application, this workload reduction may be further enhanced and will have a considerable impact regarding cost saving comparing to the traditional approach.
Recommended publications
  • FPT Software Enhancing Code Quality and Security with Synopsys AST Solutions
    FPT Software Enhancing Code Quality and Security with Synopsys AST Solutions The Challenge Improving the Quality and Security of Software Projects FPT Software offers full lifecycle services for its customers, including software design, development, migration and modernization. Providing software components for clients’ systems often means working with legacy code and architectures not originally designed for modern interconnected environments. These must be stringently tested Company Overview for the quality and security needed by today’s modern applications. FPT Software is part of FPT “Winning clients’ trust is our utmost priority, and we constantly focus on enhancing Corporation, a technology and IT code security,” says Do Van Khac, Chief Delivery Officer and Executive Vice President at services provider headquartered FPT Software. “We often encounter risks stemming from incompatible legacy code and in Vietnam with nearly USD $2 architectures, which led to increased costs in remediation. We took the path of looking billion in revenue. A pioneer in into tools to improve code quality and security as early as possible in the development digital transformation, FPT delivers lifecycle.” world-class solutions services The Delivery Command Center (DCC) at FPT performed a detailed evaluation of several in smart factory technologies, static analysis solutions; ultimately determining that Coverity Static Analysis was the digital platforms, robotic process right choice for them. By helping to identify and fix software problems early in the automation, artificial intelligence, development process, Coverity not only accelerates FPT code reviews to improve code Internet of Things, mobility, cloud, quality and security but also helps FPT reduce the need and cost of addressing those managed services, testing, and problems later.
    [Show full text]
  • Offshore Development Services Bdo Digital & Fpt Software
    Software OFFSHORE DEVELOPMENT SERVICES BDO DIGITAL & FPT SOFTWARE CONNECT. CLEVER. IMPACT. CONFIDENTIAL © 2020 BDO Software BDO Belgium PARTNERS & STAFF (Average FTE’S) 88 559 96 = 743 2018 2018 2018 2018 80 475 102 657 PARTNERS PROFESSIONAL STAFF SUPPORT STAFF TOTAL 26.4 22.8 37.4 17.6 104.2 MillionEUR MillionEUR MillionEUR MillionEUR = Million EUR 2018 2018 2018 2018 24.9 21.6 26.9 17.3 2018 90.7 AUDIT & ASSURANCE TAX & LEGAL ADVISORY ACCOUNTING & REPORTING TOTAL 140,259 OTHER STAFF RELATED INFORMATION EUR 85% 782 2018 2018 2018 138,052 EUR 84% 732 REVENUE PER STAFF STAFFRETENTION NUMBER OF STAFF OCTOBER 1ST (HEADCOUNT) 2 CONFIDENTIAL © 2020 BDO Software BDO Advisory Our product lines and services Corporate Strategy & Interim People Digital Risk Finance Organisation Management Valuation Strategy planning Leadership & talent Technology & CIO Internal Audit Interim & execution management advisory services Management Transaction Services Forensic Advisory Organisational Change Enterprise solutions Services M&A excellence management Emerging Risk Management Financing Advisory Process Excellence Training & technologies & Internal Control & Modeling Coaching Project & App development portfolio HR tools & management analytics Advanced analytics Service design & Mobility Digital industry customer journey & society HR & Pay Pre/post-merger platform Security & audit integration services 3 CONFIDENTIAL © 2020 BDO Software BDO DIGITAL Expertise Typical products Technology & IT Strategy & Roadmap | Governance, Project & Ptf. Management | Cloud Strategy & Migration CIO advisory services Services | ... Enterprise solutions Oracle Netsuite | Globis | CA Project Portfolio Management | Keyedln | IonProjects | ... Emerging technologies Blockchain & Data Sharing Platforms | Robotic Process Automation | Internet of Things | ... App development Systems Integration & API Management | Mobile Apps | ... Advanced analytics Data Mining, Machine Learning & AI | Process Mining / Intelligence | Reporting & BI | ..
    [Show full text]
  • Powering Smarter Manufacturing
    Case Study Powering Smarter Manufacturing Domain Manufacturing Challenges • Lack of a master digital system • Lack of real-time information • Inefficient and costly paper-based production line Solution Transform the factory via akaMES: • Integrate equipment and machines with sensors, bar codes, QR codes, RFID (Radio Frequency Identification) to record data in real-time • Develop a customized manufacturing execution system to monitor and manage collected data Benefits • 90% of hard copies were replaced by e-document, overhead expenses decreased by 8% • A bird’s eye view and real-time visibility into the business operations for users Content Abstract 3 The Painful Paper-based Production Line 4 Getting to the Core of Problems 5 A Bespoke Manufacturing System 6 Operation at Ease 7 Powering smarter manufacturing 2 Abstract As technology has played a growingly massive role in any aspect of life, more and more manufacturers are looking for IT solutions that can help them remain or achieve competitive advantages. Those who lag behind on the technology race may risk being out of the game permanently. How to streamline traditional, manual stages of the production line has long been the quest for many enterprises. This case study specifies the journey of a leading electronics manufacturer in leveraging digitalization to modernize and optimize its business performance. Powering smarter manufacturing 3 The Painful Paper-based Production Line • The quality of information was not ensured • Archiving proved to be a big hassle Company S has been one of the largest multinational corporations in producing and assembling electronic parts and accessories for decades. With over 20 facilities in 6 countries around the world, the company designs and builds millions of items a year, making it one of the largest and most well-known electronic manufacturers in the global market.
    [Show full text]
  • Fpt Software
    FPT SOFTWARE I. COMPANY PROFILE: Address: FPT Cau Giay Building, 17 Duy Tan, Dich Vong Hau Ward, Cau Giay District, Hanoi, Vietnam Established year: 1999 Tax code: 0101601092 Tel : +84 24 3768 9048 Fax : +84 24 3768 9049 E–mail : [email protected] Website : www.fpt-software.com Type of Company Limited Organization: Type of business: Service Business Industry : IT & Software Exports: $230 Million (FY 2016) Employees: 11,000 II. GENERAL INFORMATION: FPT Software is the global leading technology, outsourcing and IT services group headquarted in Vietnam with over 10,000 employees. Qualified with CMMI Level 5 & ISO 27001:2013, FPT Software delivers world-class services in Analytics, IoT, Mobility, Cloud, Product Engineering Services, Platform Modernization, Application Development & Management Service, and BPO services globally from 14 countries in 4 continents, including the United States, Japan, Germany, South Korea and Australia. With focus on R&D activities to drive innovation and improve service quality, the company has been serving over 450 customers worldwide, of which over 50 are Fortune 500s, including AT&T, Dupont and UPS. FPT Software has proven to have in-depth expertise in various industries: banking finance, automotive, media & entertainment, logistics & transportation, supply chain, healthcare among others. 1 The company has been strategic partners with various industry leaders namely AWS, Microsoft, GE, Siemens, AUTOSAR, SAP, IBM, Cisco. Company Service Portfolio III. SUGGESTED COLLABORATION: Finding partners of various industries who wish to improve their IT & software systems and adopt Digital Transformation to optimize business models; Finding technology companies to co-develop innovative solutions by leveraging skillful workforce from both FPT and technology partners; 2 .
    [Show full text]
  • PAID INTERNSHIP PROGRAM at the LEADING IT COMPANY in SOUTH EAST ASIA’S MOST CHARMING COUNTRY About the Program Coordinator CONTENT FPT Corporation
    PAID INTERNSHIP PROGRAM AT THE LEADING IT COMPANY IN SOUTH EAST ASIA’S MOST CHARMING COUNTRY About the program coordinator CONTENT FPT Corporation .........................................................3 FPT University .............................................................4 Program overview & offers ................................................5 Prestigious & enjoyable workplace ..............................6 Extensive learning & career opportunities ....................7 Exclusive financial support ..........................................8 Application & Contact .....................................................9 FPT – Powering Digital Transformation The leading ICTThe 31Corporation Leading years of ICT restless in CorporationVietnam transformation in Vietnam FPT Corporation Founded 1988 Public listed Dec 2006 on VNSE FPT Software ~USD 1.021 billion – excluding major investment in Revenue Distribution & Retail (2018) FPT Information System FPT Software: 370 Million consolidated (2018) FPT Telecom No. of Employees 34,000+ ~ FPT Software: 15,000+ Development Centers Vietnam, Philippines, Myanmar, Slovakia FPT Online Clients 600+ clients with 100 Fortune 500s FPT Education R&D Investment 5% profit before tax FPT – Leading ICT Corporation in Vietnam 01 02 TECHNOLOGY SECTOR TELECOMMUNICATIONS SECTOR • #1 in Revenue and Workforce; • To p 3 in broadband internet services, covering 59/63 • ~15,000+ Tech experts & engineers; provinces; • #1 SI provider in Vietnam and expanding to developing • 700Gbps international bandwidth; 04
    [Show full text]
  • FINANCIAL PERFORMANCE HIGHLIGHTS in NINE MONTHS of 2017 in the First Nine Months of 2017, FPT's Revenue and Profits Indicators All Showed Growth As Planned
    FPT CORPORATION THE SOCIALIST REPUBLIC OF VIETNAM Independence - Freedom - Happiness No.: �J/FPT - FCC th Hanoi, October 24 , 2017 FINANCIAL PERFORMANCE HIGHLIGHTS IN NINE MONTHS OF 2017 In the first nine months of 2017, FPT's revenue and profits indicators all showed growth as planned. Consolidated revenue wasVND 31,131 billion, up 9% Yo Y. Profit beforetax (PBT) was VND 2,308 billion, increasing by 15% Yo Y and equal to 99% of the YTD target. Profit after tax (PAT) was VND 1,955 billion, up 12% YoY. Year-to-date PAT attributable to parent company's shareholders was VND 1,507 billion, up 13% YoY. Year-to-date EPS was VND 2,845 in the nine-month period, up 13% YoY. FPT's earnings growth in nine-month period continued to be driven by the two core business sectors including Technology and Telecom, which accounted for73% of the consolidated PBT of the group. Specifically, the Technology and Telecom sectors' profit before tax were up 16% and 10% YoY, respectively. FPT's overseas markets recorded revenue ofVND 4,879 billion, up 17% YoY, and PBT ofVND 741 billion, up 23% YoY, accounting for nearly one-third of the consolidated PBT. SUMMARY OF NOTABLE EVENTS •!• Synnex invests to own 47% of FPT Trading: FPT Corporation has signed an investment agreement with the strategic partner Synnex Technology International Corporation (Synnex) - A USD-33-billion-revenue corporation and being the world's third largest distribution group in IT, telecom and JC component distribution. Accordingly, Synnex will invest to own 47% of charter capital of FPT Trading Company Limited (FPT Trading).
    [Show full text]
  • FPT Corporation ADD (Previously NOT RATED) Riding the Digital Transformation Wave Consensus Ratings*: Buy 12 Hold 0 Sell 0
    Company Note Conglomerate │ Vietnam │ December 11, 2019 Shariah Compliant Insert Insert Vietnam FPT Corporation ADD (previously NOT RATED) Riding the digital transformation wave Consensus ratings*: Buy 12 Hold 0 Sell 0 Current price: VND55,500 ■ FPT is Vietnam’s largest tech company and is involved in the IT services, Target price: VND74,000 telecom and education businesses. Up/downside: 33.3% ■ We forecast an FY19-21F net profit CAGR of 12.1% on the back of an CGS-CIMB / Consensus: 0.2% increasing Dx trend and higher proportion of patented software products. Reuters: FPT.HM ■ Initiate coverage with an Add rating and target price of VND74,000. Bloomberg: FPT VN Market cap: US$1,625m FPT is moving up the value chain VND37,644,336m Initially, FPT was a pure service-based IT company with few patented software products. Average daily turnover: US$4.25m This could partially explain why FPT’s P/E has been consistently lower than the Vietnam VND97,815m Stock Index (VN-Index)’s P/E over the past 10 years, in our view. In late-2017, however, Current shares o/s: 613.636m Free float: 78.9% the company embarked on a strategy to shift its focus to technology through the *Source: Bloomberg divestment of the digital retail and distribution businesses. Additionally, FPT ramped up R&D activity to strengthen its product line-up with patented software products. The Key changes in this note valuation gap between FPT and VN-Index has begun to narrow but we believe this local N/A tech giant deserves to trade at a premium over the index.
    [Show full text]