Search This Blog


Saving More lives with AI

Samsung SDS is an abbreviation for Samsung data services, and it is one of about The 20 businesses that comprise the Samsung Group: famous for producing mobile phones, television sets, laptop computers, semiconductors, hard discs, and home electronics refrigerators, dishwashers, and even operating systems in heavy industry, shipbuilding.

SDS's primary product is the operation of data centres, in which they also participate run the Brightics AI Accelerator platform for automated machine learning (AutoML); allowing users to develop AI models much more easily more effectively, precisely, and quickly than ever before

Vice President of Dr Patrick Bangert, Samsung SDSA's Artificial Intelligence, oversees AI engineering and AI sciences teams. The AI engineering team creates. Brightens AI Accelerator provided automated machine learning and distributed training understanding how to accelerate the construction of an AI model. The AI sciences team develops models and expert consulting services. AutoLabel automatically labels prepare datasets for AI by annotating their modelling. They provide the whole range of AI model creation supported by cutting-edge technology and human knowledge. The AI team looks for initiatives both within and outside. They are retaining their position outside the Samsung Group's computer infrastructure and protecting a worldwide presence.

The Human Advantages of Automation of imaging procedure

Samsung SDS has placed a strong emphasis on the healthcare industry, where AI has been used image process automation in regions as in radiology Bangert asserts in no." The medical system" uses ambiguous terms: The planet is overburdened. There aren't any. There are enough physicians to go around. They are greatly influenced by demand due to processes, bureaucracy, and charging procedures. The length of time that any doctor may devote time to genuine patients. Human patients constitute a small proportion of their whole population. Time. A doctor's advice may be incorrect. Around 30% to 35 % of the time, complicated illnesses – identified from complex medical images - are often used as either overlooked or misdiagnosed.

A second opinion is required. already worth a billion dollars today's market." Second thoughts generally originate from other people. Artificial Intelligence offers a neutral third party and a very neutral second. Obtaining a second or third opinion on any method for diagnosing AI eliminates the possibility of any one doctor having a specific level of experience, since

The AI model can handle much more data. Instances and never forgets anything.

The director, Ricky Datta of artificial intelligence engineering at SDS from Samsung, observes;

"Typically, AI models attain precision in the upper 90 percentile, Up to 99.9% of the time in certain circumstances, precision depending on the amount what information is accessible and how If it is of high quality is. By automating these procedures We merely use processes accelerate everything This would be advantageous. greatly reduce the expenditures incurred by both As providers and patient care expenses enhancing diagnostic accuracy and speed increase the total time from the first discomfort point until the beginning of the therapy cycle."

This is true in all medical fields. MRI and CT scan imaging, rather than X-ray Converting ultrasound pictures to standard photographs of the skin Zakia Rahman, clinical assistant professor of dermatology at Stanford University School of Medicine Dermatology, according to the Department of Medicine, "represents two significant new business model boundaries artificial intelligence-powered Patients can snap images of their skin using their phones and acquire immediate, trustworthy diagnoses for any state, and individuals may make use of photographs of to compare their healthy features and bodies for comparing them to society's beauty standards precise cosmetics for a better appearance while not looks unusual The most significant impediment to each include a vast carefully marked

The ultimate AI toolbox from Samsung: 

Brightens Artificial Intelligence Accelerator

The crowning achievement of Samsung's AI efforts is the Brightics AI AutoLabel feature Platform for acceleration. Hankyu Moon, Dr., the team leader behind AutoLabel, reasons this toolbox is so valuable in three sections "First and foremost, in the case of imaging, for An annotation, for example, is often a guidebook. Drawing a circle around something thing is significant, as is categorizing. It has a name. This may be seen in street scenes. Where we may draw the line with autonomous cars an outline around individuals to indicate 'ok, this is it's a person who represents an impediment that the 'The automobile must not collide.'" He describes how these annotations are created. Fast by the AutoLabel feature, which essentially arranges the photographs in the correct order manner with which the information is provided. The trick is that just a handful, generally 10%, of Almost all of the photos in a piece of dataset information - yet locating it is difficult.

The second section, connected to the Moon AutoML Brightics AI Accelerator platform for feature research engineering, entails pre-processing the data in such a manner as to reveal those characteristics of the data that are most important informed, as well as picking the appropriate situations model.

According to Moon, artificial intelligence is a catch-all word for various unique methodologies and model kinds, such as Support vector machines, neural networks, Decision trees or a random forest. It's a situation. of selecting the best one.

Moon said that the third section, AI's role in healthcare imaging, is quite exciting.

Tweaking parameters: "The algorithm that parameters are used to train one of these models by itself. They are often determined by a human, leading to a trial-and-error process of fine-tuning these things properly. AutoML optimizes them for you. automatically."

When it comes to AI accelerators, Another strategy is dispersed training. Datta explains to Samsung: "We use more than one graphics processing unit (GPU). We may employ to assist speed up the procedure. Many of these GPU processors, spread concurrently across multiple to use computers to complete a single training objective." As the go-to toolbox, the AI accelerator The goal of Samsung SDS is to work on all platforms. The efficiency, labelling, and tweaking of the models and distributed training concurrently, such They can "execute the artificial intelligence." process much more swiftly and effectively and arrive at a much more accurate result. Datta concluded, "model in the end."

Healthcare Imaging Difficulties

Nasim Eftekhari, applied AI director Leveraging data science at City of Hope, a world-class cancer research and treatment facility "All," says an organization near Los Angeles. The supervised models that we employ nowadays are training on labelled data. Regardless, Labeling data for training purposes in the industry is generally the most costly component of any AI-based solution. It takes a long time, and since physicians and healthcare are pricey, These photographs must be annotated by specialists. And each picture takes hours and hours to create in terms of the time the City of Hope wants to investigate Using Samsung's active learning in picture auto labelling. There is no doubt about that. About the importance of a deep understanding of pathology picture auto-labelling, We are pleased to investigate the concept of a system that can learn incrementally from "A doctor takes 20 to 30 minutes." minutes to doodle a comment on top of a single picture To train an AI algorithm, to recognize something significant, There are hundreds of such annotated photographs. Required. The procedure is the most time-consuming and costly of the complete course. Workflow for artificial intelligence

It is the most significant impediment that the Brightics AI's AutoLabel feature Accelerator is used to overcome obstacles. Active learning strategy to accelerate that procedure by 90%, hence it will save money the amount of human labour reduced from 100% to 10% - a significant benefit, generally in the tens of millions of dollars for each model.

Dan Waters faces a new task. Business development director and similar goods at Samsung SDS to his residence in the United States, The FDA is the Food and Drug Administration (FDA). Any physical gadget used must be approved. In the medical field, Waters claims, "Because government agencies" are not as sophisticated in their handling of obtaining a physical kind of artificial intelligence, It might take a long time to get a gadget certified. The procedures and staff in place are merely unfamiliar with this new technology. As far as they are concerned, AI is technology. Is still in its early stages."

AI is used outside of healthcare.

"The single greatest problem is climate change." of humanity," Bangert claims.

"AI has a significant role to play in restricting the increase of greenhouse gases. This has and a lot to do with how well we can run things in industries, automobiles, aircraft, and people's homes houses," he went on. Samsung's consumer electronics and semiconductors Home automation gadgets may assist you. Its energy efficiency lowers any individual's entire carbon footprint house or structure. Data centres, particularly AI Workloads, are often blamed for worsening the problem. The contemporary world's climatic effect, yet this is short-sighted. "Other carbon emission sources" Physical world variables are avoided. Because of artificial intelligence, Consider the significant application. You are purchasing on the internet. You visit online, buy something, and the post office delivers it. That is much more. Efficient in terms of greenhouse gas emissions than if you drive your SUV by yourself

"AI can go 20 kilometres to the next mall." to change the world in healthcare diagnosis and therapy as well as clerical procedures, new employment creation and letting current employees improve efficiency."


Vice President of AI,