What is Machine Learning?

As introduced in the first episode of the All About AI series, machine learning is an algorithm that autonomously learns patterns from data to make predictions. It has become the dominant methodology in AI, especially with the explosive growth of data in recent years. In contrast, traditional AI models required humans to explicitly program rules and logic. While these models worked well for problems with clear, defined rules, such as simple board games, they had limitations when dealing with complex rules and datasets. For example, consider the task of developing an AI image recognition application to identify cats. This would involve steps such as processing countless pixels, interpreting color data, and establishing rules to distinguish a cat’s patterns, highlighting the challenges of such projects.

Figure 1. Machine learning models can distinguish dogs from cats by learning from large datasets of labeled images

Machine learning involves training a system to discover complex structures and patterns within data, enabling it to learn these patterns independently and make predictions based on new data. As an example, imagine if the aforementioned AI image recognition software to detect cats is created using machine learning. For this process, various photos are collected to train the algorithm, allowing it to learn how to identify cats on its own.

Types and Characteristics of Machine Learning Algorithms

Machine learning algorithms extract data from underlying probability distributions¹ in the real world to train models for tackling specific problems. These algorithms can be broadly categorized into three types, each with distinct characteristics and applications based on the type of problem.

¹Probability distribution: A mathematical model that describes how data is distributed, enabling the identification of patterns and structures within the data. This signals the probability that various outcomes of a given system will occur under different conditions.

1) Supervised learning: This method involves training a model using inputs paired with correct outputs, or labels, and then analyzing these training pairs to predict outcomes for new data. As an example, imagine developing an AI application to predict a person’s gender from a photo. Here, the photo is the input, and the gender is the label. The model learns the characteristics that distinguish gender and applies this knowledge to predict the gender of individuals in new photos. Supervised learning can be further divided into two types depending on the nature of the labels:

• • Classification: This algorithm assigns data to specific categories using labels. For example, determining whether a photo contains a dog or distinguishing letters in handwritten text are classification tasks. In these scenarios, the data is assigned to a specific category and labeled accordingly.
  • Regression: This technique uses labels for continuous values. For example, predicting house prices based on factors such as size and location, or forecasting temperatures from current weather data, are typical regression problems. In these cases, the goal is to predict numerical values as accurately as possible.

2) Unsupervised learning: As the name suggests, unsupervised learning differs from supervised learning in that it learns from data without explicit “supervision”, meaning there are no labels provided. Instead, unsupervised learning focuses on understanding and learning the characteristics of the data’s probability distribution. Key techniques include:

• • • Clustering: This method groups data with similar characteristics to identify hidden patterns in probability distributions. For example, in a real-world semiconductor manufacturing process, applying a clustering algorithm to images of defective wafers² revealed that the defects could be categorized into several types based on their underlying causes.
    • Dimensionality reduction: This technique simplifies high-dimensional data—datasets with numerous features or variables—while retaining the most important information. Thus, it aids in data analysis and visualization. A common example is principal component analysis (PCA), which removes unnecessary information such as noise—irrelevant or random variations.

²Wafer: Generally made from silicon, wafers act as the substrate, or foundation, for semiconductor devices.

Generative AI, which has gained significant traction recently, generally fits into the unsupervised learning category as it learns probability distributions from data to generate new data. For example, ChatGPT learns the probability distribution of natural language to predict the next word in a given text. However, since generative AI often uses supervised learning techniques during training, there is debate about whether it should be considered purely unsupervised learning.

3) Reinforcement learning: This type of learning focuses on training a model to maximize rewards through interaction with its environment. It is particularly effective for tasks requiring sequential decision-making. Thus, it is widely used in robotics to help robots find optimal paths while avoiding obstacles, as well as in autonomous driving and AI gaming. Recently, reinforcement learning from human feedback (RLHF)³ has garnered significant attention for ensuring large language models (LLMs)⁴ such as ChatGPT, enabling them to generate responses which better align with human preferences.

³Reinforcement learning from human feedback (RLHF): A method in which a model learns from “rewards” given based on human feedback. By incorporating human evaluations of its outputs, the model adjusts its behavior to generate responses that better match human preferences.
⁴Large language models (LLM): Advanced AI systems trained on vast amounts of text data to understand and generate human-like text based on the context they are given.

Figure 2. A video clip showing an AI playing a brick-breaking game. In this classic example of reinforcement learning, the AI is given the instruction to break more bricks to improve its score and learns to do so on its own.

Evaluating Machine Learning Performance

The ultimate goal of machine learning is to perform effectively with new, unseen data in real-world situations. In other words, it is important for the model to have generalization capabilities. To this end, it is essential to accurately evaluate and verify the model’s performance, which typically involves the following steps.

1) Choosing Performance Metrics
The choice of performance metrics depends on the type of problem being addressed. In classification problems, common performance metrics include:

• • Accuracy: Refers to the proportion of correct predictions. For example, if a medical test correctly diagnoses 95 out of 100 cases, the accuracy rate is 95%. However, a meaningful accuracy assessment requires a balanced dataset. If 95 out of 100 samples are negative and only five are positive, a model predicting all samples as negative would still achieve 95% accuracy. This high level of accuracy can be misleading, as the model would fail to identify any positive samples.
  • Precision and recall: Precision measures the proportion of actual positives among predicted positives, while recall measures the proportion of correctly predicted positives among actual positives. These metrics have a trade-off relationship, so it is essential to optimize the model by considering the balance and objectives between precision and recall. For example, in medical testing, increasing recall is considered crucial to detect as many cases of a condition as possible, while precision might be more important for tasks such as spam mail filtering. To address this trade-off problem, the F1 score⁵ is used to evaluate the balance between precision and recall.

For regression problems, performance is typically assessed using metrics such as mean squared error (MSE)⁶ , root mean squared error (RMSE)⁷ , and mean absolute error (MAE)⁸.

⁵F1 score: The harmonic mean of precision and recall, useful for evaluating a model’s classification performance when class imbalance exists. It ranges from 0 to 1, with higher values indicating better performance.
⁶Mean squared error (MSE): The average of the squared differences between predicted and actual values.
⁷Root mean squared error (RMSE): The square root of the MSE, providing error measurement in the same units as the actual values.
⁸Mean absolute error (MAE): The average of the absolute differences between predicted and actual values.

2) Performance Evaluation Methods

Figure 3. A visual representation of the train-test split method and cross-validation techniques used in machine learning

To evaluate machine learning models, data is usually split into training and testing sets. This helps assess how well the model generalizes new data.

• • Train-test split: As one of the simplest methods, it involves dividing data into a training set and a test set. The model is trained on the training set, and then its predictive performance is evaluated on the test set to gauge generalization performance. Typically, 70–80% of the entire data is used for training, with the remainder reserved for testing.
  • Cross-validation: This model divides the data into so-called K folds, in which K refers to the number of data subsets, or folds. One fold is used for testing, while the other K-1 folds are used to train the model. This process is repeated K times, and the average performance is calculated. Although cross-validation is common in traditional machine learning, it is time-intensive, making the train-test split the preferred method in deep learning.

3) Understanding Model Performance
The results from these evaluation methods provide critical feedback for improving model performance. However, two common phenomena often arise:

• • Underfitting: This occurs when a model is too simple to learn the underlying patterns in the data, resulting in poor performance on both training and testing sets. For example, consider a regression problem where the actual data follows a quadratic function but the predictive model is set as a linear function. The model may in this case be unable to identify patterns, meaning it has low expressivity as it cannot express complex functions, possibly leading to underfitting.
  • Overfitting: This occurs when a model is too complex and learns both the data’s patterns and its noise, performing well on training data but poorly on test or new data. To address overfitting and better assess generalization, techniques like cross-validation can be used. Evaluating the model’s performance across various data splits allows for a more accurate assessment of overfitting and helps in selecting the appropriate level of model complexity.

To build a model with strong generalization ability, it is widely recognized that a balance must be reached between underfitting and overfitting through methods such as regularization⁹. Interestingly, recent research in deep learning has revealed a double descent¹⁰ phenomenon, where increasing the model size after initial overfitting does not actually worsen overfitting but can improve generalization performance. This discovery has sparked significant research interest in these areas.

⁹Regularization: A method to prevent overfitting by limiting model complexity or adding penalty terms.
¹⁰Double descent: A phenomenon in which model performance worsens with increasing size up to a point, then improves beyond a certain size. This challenges traditional views on overfitting in deep learning, though it remains theoretically unexplained as it is a newly overserved phenomenon in deep learning.

This concludes the All About AI series, which has provided an overview of the basics of AI and machine learning. Stay tuned to the newsroom for the latest updates on AI memory technologies driving innovation in the industry.

<Other articles from this series>

[All About AI] The Origins, Evolution & Future of AI

The post [All About AI] A Guide to Machine Learning Fundamentals first appeared on SK hynix Newsroom.

AI-powered robots that walk, talk, and think like humans have long been a staple of sci-fi comics and movies. However, AI and robotics are no longer merely works of fiction—they have become a reality. Now that AI is here and transforming people’s lives, it is prudent to look back and consider AI’s origins, the milestones which have shaped the technology’s evolution, and consider what the future might hold.

From the Turing Test to Machine Learning: AI’s Early Beginnings

Figure 1. An overview of AI’s evolution through the decades from the 1950s to the 2020s

The birth of AI can be traced back to the 1950s. In 1950, British mathematician Alan Turing proposed that machines could “think,” introducing what is now known as the “Turing test” to evaluate this capability. This is widely recognized to be the first study to present the concept of AI. In 1956, the Dartmouth Summer Research Project on Artificial Intelligence formally introduced the term “AI” to the wider world for the first time. Held in the U.S. state of New Hampshire, the conference fueled further debates on whether machines could learn and evolve like humans.

During the same decade, the development of artificial neural network¹ models marked a significant milestone in computing history. In 1957, U.S. neuropsychologist Frank Rosenblatt introduced the “perceptron” model², empirically demonstrating that computers can learn and recognize patterns. This practical application built on the “neural network theory” developed in 1943 by neurophysiologists Warren McCulloch and Walter Pitts, who conceptualized nerve cell interactions into a simple computational model. Despite these early breakthroughs raising high expectations, research in the field soon stagnated due to limitations in computing power, logical framework, and data availability.

¹Neural network: A machine learning program, or model, that makes decisions in a manner similar to the human brain. It creates an adaptive system to make decisions and learn from mistakes.
²Perceptron: The simplest form of a neural network. It is a model of a single neuron that can be used for binary classification problems, enabling it to determine whether an input belongs to one class or another.

Then in the 1980s, “expert system” emerged which operated solely based on human-defined rules. These systems could make automated decisions to perform tasks such as diagnosis, categorization, and analysis in practical fields such as medicine, law, and retail. However, during this period, expert systems were limited by their reliance on rules set by humans and struggled to understand the complexities of the real world.

In the 1990s, AI evolved from following human commands to autonomously learning and discovering new rules by adopting machine learning algorithms. This became possible due to the advent of digital technology and the internet, which provided access to vast amounts of online data. At this point, AI was able to unearth new rules even humans could not discover. This period marked the start of renewed momentum for AI research, based on machine learning.

The Rise of Deep Learning: A Key Technology in AI’s Growth

Figure 2. Timeline showing advances in artificial neural networks and deep learning

While the 1990s presented opportunities for AI to grow, the journey and evolution of AI has had its share of setbacks. In 1969, early artificial neural network research hit a roadblock when it was discovered that the perceptron model could not solve nonlinear problems³, leading to a prolonged downturn in the field. However, computer scientist Geoffrey Hinton, often hailed as the “godfather of deep learning,” breathed new life into artificial neural network research with his groundbreaking ideas.

For example, in 1986, Hinton applied the backpropagation⁴ algorithm to a “multilayer perceptron” model, essentially layers of artificial neural networks, proving it could address the limitations of the initial perceptron model. This seemed to spark a revival in artificial neural networks research, but as the depth of the networks increased, issues began to emerge in the learning process and outcomes.

In 2006, Hinton introduced the “deep belief network (DBN),” which enhanced the performance of a multilayer perceptron, in his paper “A Fast Learning Algorithm for Deep Belief Nets.” By pre-training each layer through unsupervised learning⁵ and then fine-tuning the entire network, the DBN significantly improved the speed and efficiency of neural network learning—which had previously been deemed an issue. This progress paved the way for future advancements in deep learning.

³The initial perceptron model was a single-layer perceptron that could not solve nonlinear problems such as the XOR problem, which involves two input values; it outputs 0 if the two input values ​​are the same and 1 if they are different.
⁴Backpropagation: An algorithm used in neural networks to minimize errors by adjusting the weights. It works by calculating the difference between the predicted and actual values and then updating the weights in reverse order, starting from the output layer.
⁵Unsupervised Learning: A type of machine learning where the model is trained on input data without explicit labels or predefined outcomes. The goal is to discover and understand hidden structures and patterns within the data.

In 2012, deep learning made a historic leap forward when Hinton’s team won the ImageNet Large Scale Visual Recognition Challenge (ILSVRC) with their deep learning-based model, AlexNet. This triumph demonstrated deep learning’s immense power by recording an error rate of just 16.4%, surpassing the 25.8% of the previous year’s winner.

Figure 3. An Overview of ILSVRC’s Image Recognition Error Rate by Year (Kien Nguyen, Arun Ross, Iris Recognition With Off-the-Shelf CNN Features: A Deep Learning Perspective, IEEE Access, Sept. 2017 p.3)

Deep learning, a focal point of AI research, has grown rapidly since the 2010s for two primary reasons. First, advances in computer systems, including graphics processing units (GPUs), have driven AI development. Originally designed for graphics processing, GPUs can process repetitive and similar tasks in parallel. This capability enables GPUs to process data faster than central processing units (CPUs). In the 2010s, general-purpose computing on GPUs (GPGPU) emerged, enabling GPUs to be used for broader computational tasks beyond graphics rendering and allowing them to replace CPUs in some instances. The use of GPUs has further increased as they have been utilized for training artificial neural networks, accelerating the development of deep learning. Deep learning, which needs to perform iterative computations during analysis of large datasets to extract features, benefits from the parallel processing capability of GPUs.

Second, the expansion of data resources has fueled progress in deep learning. Training an artificial neural network requires vast amounts of data. In the past, data was primarily sourced from users manually inputting information into computers. However, the explosion of the internet and search engines in the 1990s exponentially increased the range of data available for processing. In the 2000s, the advent of technologies such as smartphones and the Internet of Things (IoT) contributed to the birth of the Big Data era, where real-time information flows from every corner of the globe. Deep learning algorithms use this large quantity of data for training, growing increasingly sophisticated. This data revolution has therefore set the stage for significant advancements in deep learning technology.

Figure 4. Google DeepMind’s AlphaGo – The Movie is a documentary film about the epic battle between AlphaGo and Lee Sedol on March 9, 2016

By 2016, the evolution of AI reached a dramatic turning point with the development of AlphaGo, an advanced AI program created by Google DeepMind to play the board game Go. This extraordinary AI program captivated the world when it defeated Go grandmaster Lee Sedol by an impressive 4-1 score. Combining deep learning with reinforcement learning⁶ and Monte Carlo tree search (MCTS)⁷ algorithms, AlphaGo learned to mimic human intuition, predict moves, and strategize through tens of thousands of self-played games. AlphaGo’s victory over a legendary human player signaled the beginning of a new AI era.

⁶Reinforcement Learning: A type of machine learning where an AI agent learns to make decisions by interacting with an environment. The agent receives rewards or penalties based on its actions and aims to maximize cumulative rewards over time by optimizing its strategy.
⁷Monte Carlo tree search (MCTS): A stochastic algorithm that repeatedly generates a series of random numbers to derive a numerical approximation of a function’s value. It structures the possible actions of the current situation into a search tree and uses random simulations to infer the pros and cons of each, ultimately determining the optimal course of action.

ChatGPT: The Catalyst for the Generative AI Boom

Figure 5. Generative AI explained through key AI subsets

At the close of 2022, humanity stood on the brink of a transformative leap with AI technology. OpenAI unveiled ChatGPT, powered by a type of LLM⁸ known as generative pre-trained transformer (GPT) 3.5, marking the dawn of the generative AI era. Most notably, this leap propelled AI into the creative realm, a domain once considered uniquely human. Now, generative AI can produce high-quality content across diverse formats, moving beyond traditional deep learning, which merely predicts or classifies data. Instead, generative AI, using LLMs or various image generation models such as variational autoencoders (VAEs), generative adversarial networks (GANs), and diffusion models, creates original results tailored to user needs.

⁸Large language model (LLM): Deep learning algorithms that perform a variety of natural language processing tasks by leveraging extensive data.

To provide a clearer context for the evolution of generative AI, it is essential to examine its origins and key developments. The roots of generative AI trace back to 2014, when American scientist and researcher Ian Goodfellow introduced GANs. In GANs, two neural networks engage in a continuous duel: one generates new data from a dataset, while the other network compares this new data to the original dataset to determine its authenticity. Through this iterative process, GANs produce increasingly refined and sophisticated outputs. Over time, researchers have enhanced and expanded upon this model, leading to its widespread use in applications such as image generation and transformation.

In 2017, the natural language processing (NLP)⁹ model “transformer” was introduced. This model considers the relationships between data as key variables. By focusing more attention to certain information, transformers can learn complex data patterns and relationships between data, capturing essential details to produce higher quality results. This advancement transformed NLP tasks such as language comprehension, machine translation, and conversational systems, leading to the development of LLMs such as the aforementioned GPT.

⁹Natural language processing (NLP): A subfield of AI that uses algorithms to analyze and process natural language data. By examining syntactic structures, semantic relationships, and contextual patterns, NLP systems can perform tasks such as language translation.

First released in 2018, GPTs have rapidly advanced in performance by expanding their parameters and training on data every year. By 2022, OpenAI’s chatbot ChatGPT, powered by GPT-3.5, completely changed the paradigm of AI. ChatGPT, with its exceptional ability to understand user context, deliver relevant responses, and handle diverse queries, quickly gained traction. Within a week of its launch, it drew over 1 million users and attracted more than 100 million active users within two months.

The rapid advancements in AI culminated in a major technological leap forward in 2023 with the launch of GPT-4 by OpenAI. This new model is built on a dataset roughly 500 times larger than that of GPT-3.5. GPT-4, now considered a Large Multimodal Model (LMM)¹⁰, can simultaneously process diverse formats of input data, including images, audio, and video, expanding far beyond its text-only predecessors. In 2024, OpenAI introduced GPT-4o, an enhanced model offering faster, more efficient processing of text, voice, and images. Capitalizing on the generative AI boom triggered by ChatGPT, companies have rolled out diverse services. For example, Google’s Gemini can simultaneously recognize and understand text, images, and audio; Meta’s SAM accurately identifies and isolates objects in images; and OpenAI’s Sora generates videos from text prompts.

¹⁰Large Multimodal Model (LMM): A deep learning algorithm that can handle many types of data, including images, audio, and more, not just text.

The generative AI market is only beginning to unleash its potential. According to a report from the global market research firm International Data Corporation (IDC), the market is set to be worth 40.1 billion USD in 2024—2.7 times larger than the previous year. Looking ahead, the market is expected to continue its growth each year and reach 151.1 billion USD by 2027. As generative AI evolves, its influence will extend beyond software to various formats including hardware and internet services. The world can expect a leap in capabilities and a push towards greater accessibility, making cutting-edge AI technology available to an ever-growing audience.

AI’s Impact on Revolutionizing Today and Redefining Tomorrow

Just as Google search revolutionized the early 2000s and mobile social media reshaped the 2010s, AI is now driving transformative changes across society. The pace of this technological advancement is unprecedented, and the challenges and concerns of humanity are growing along with it.

So what is the “next generative AI”? The most notable technology around today is perhaps on-device AI. Unlike traditional AI that relies on large cloud servers to pull data to edge devices, on-device AI operates directly on electronic devices such as smartphones and PCs through integrated AI chipsets and smaller LLMs (sLLMs). This shift promises to enhance security, conserve resources, and deliver more personalized AI experiences.

Figure 6. Cloud-based AI vs on-device AI structures

AI will seamlessly integrate into an increasing number of devices, continuously evolving in form and function. Thus, innovations that once seemed like science fiction are becoming reality. For instance, in 2023, U.S. startup Humane launched the AI Pin, a wearable device with a laser-ink display that projects a menu onto the user’s palm. At CES 2024, Rabbit’s R1 and Brilliant Labs’ Frame showcased their own cutting-edge wearable AI technology. Meanwhile, mixed reality (MR) headsets, like Apple’s Vision Pro and Meta’s Quest, are pushing beyond traditional virtual reality (VR) and metaverse experiences, opening up new markets.

However, as technology races forward, it not only creates new opportunities but also brings about social challenges. The rapid rise of AI has sparked concerns about society’s ability to keep up with these advancements. In particular, AI’s potential misuse and impact on real-world issues has heightened these fears. Sophisticated AI-generated content, such as deepfake videos and manipulated images, creates fake news and disrupts society. Recently, concerns about fake content have intensified in many countries ahead of major elections, including the 2024 U.S. presidential election.

Figure 7. Social anxiety and disruption due to deepfake technology portrayed by DALL-E, a generative AI platform

There are also risks associated with the development and use of AI. As generative AI crawls and merges publicly available web contents to train its AI models, there are concerns about plagiarism. Moreover, copyright disputes can arise from creating content using similar prompts with the same generative AI program. The potential for AI to shift from enhancing productivity to replacing jobs and disrupting the labor market presents a troubling reality for some as well.

AI has created a world beyond human imagination. As this new world unfolds, it is crucial to prepare for the changes ahead. Addressing this new era involves thoughtful planning and social discussion. These action items first require a deep understanding of AI’s potential and implications, which will be provided throughout the All About AI series.

The post [All About AI] The Origins, Evolution & Future of AI first appeared on SK hynix Newsroom.

“A World without Barriers”, the theme of the 2022 International Translation Day (September 30), has long been our hopeful vision for the future world. For years, technologies like AI (Artificial Intelligence) have been supporting our reach for the wildest dreams. With the rise of AI translation, a world without language barriers seems closer than ever before.

The Rising Demand for Translation

Under the wave of globalization, different countries are now extremely connected in various fields such as economy, trades, and culture. As effective communication and mutual understanding are fundamental for international collaborations, breaking language barriers to accurately reach consensuses has become an urgent issue. As a matter of fact, a study shows that above other culture factors such as value differences or stereotypes, speaking different languages is still the primary challenge for effective cross-culture communication².  As a result, the global language market is constantly expanding, expected to reach 57.7 billion US Dollars this year³.

While English, widely used and learnt around the world, has established itself as a world language, about 40% of the world’s languages are on the verge of extinction. As the language diversity is indispensable for the diversity of culture, to prevent the loss of minority cultures, we need to hear the voices of endangered languages. Translation has thus become our key access to reach these cultures.

So far, translation services can be roughly categorized into two types:  human translation and machine translation. In the pursuit of “faithfulness, expressiveness and elegance”, human translation can usually offer higher-quality outcomes. However, with closer global communication comes the exponential growth of translation workload, making it difficult for human translation to fully take on the challenge. Within the massive data to be translated, there are also lots of redundant, miscellaneous information, for which the usage of human translation can be quite wasteful due to the high labor cost.  In addition, human translators’ own interpretation and style might have an impact on the translation outcome. Therefore, since its appearance in mid-20^th century, machine translation has gradually become a common tool for our daily translation needs.

Figure 1. Global Machine Translation Market Size

Development of Machine Translation

The development of machine translation can be divided into three stages: Rule-based Machine Translation (RBMT), Statistical Machine Translation (SMT), and Neural Machine Translation (NMT).

Figure 2. Development of Machine Translation

Among them, neural machine translation adopts an end-to-end encoder-decoder structure, without applying preset translation rules, splitting sentences, or translating word-by-word. It directly decodes the source text, globally processing the input and output content. Decades of neural networks development has lay down the foundation for neural machine translation to achieve extremely rapid growth.

Figure 3. NMT Optimizes the Translation Process

From 2015 to 2016, Baidu and Google successively released their own self-developed online NMT systems with the most advanced training technology at that time, kickstarting the era of NMT. Since then, many other major Internet and ICT companies have also started venturing into the NMT arena, each integrating its own system with corporate visions or products/services.

For example, the “No Language Left Behind (NLLB)” project by Meta aims to help billions of people around the world translate over 200 languages with high quality. In July this year, the company announced a plan to build an open-source AI model for “NLLB”, which includes 50+ billion parameters, and is trained by an AI supercomputer, estimated to perform over 25 billion translations per day⁴. On the other hand, NVIDIA’s own Maxine Software Development Kit (SDK) is designed to provide better real-time communication experiences. It offers high-quality real-time audio translation through the AI-driven SDK. Other than that, with its augmented reality (AR) SDK, other interactive features such as face tracking and eye contact can be introduced to bring smoother and more intuitive communication to video-calls⁵.

Memory Upgrades for the Diversity & Prosperity of All Languages

The evolution of machine translation puts forward both higher requirements and bigger motivation for the development of computer technology. Since the rise of SMT, the construction and continuous expansion for text corpus has brought up another challenge for data storage.

With the development of ICT, the amount of data generated by everyday communications has become immeasurable: according to IBM’s statistics, 2.5 exabytes of data are generated every day⁶. And very likely, this amount will be further increased with the progress of civilization, as well as the diversification and complication of technology applications. To ensure the effective cross-culture communication and protect culture diversity, information and communication technologies should provide powerful timely supports. Performances of semiconductor chips, such as data transmission speed and reading/writing speed are becoming essential factors for considerations.

Figure 4. World’s first 238-layer 512Gb TLC 4D NAND developed by SK hynix

In the era of AI and big data, fast data access has become a primary technological requirement to maximize the use of massive data corpus for more effective machine translation. The development of flash memory technologies therefore plays an important role. This August, SK hynix announced the successful development of the world’s first 238-layer 512Gb TLC 4D NAND, which is expected to begin mass production in the first half of 2023. The data-transfer speed of the 238-layer product is 2.4Gb per second, a 50% increase from the previous generation. This 4D product have a smaller cell area per unit compared with 3D, leading to higher production efficiency. It will be provided for high-capacity SSDs for servers in the future, potentially helping the operation of massive text corpora.

Meanwhile for NMT, the application of deep learning is indispensable. As a highly complex form of machine learning, deep learning aims to enable machines to have learning and analytical capabilities like humans do. Running the data to be learnt on computing devices, it eventually comes up with an optimal result through billions of neural network operations and adjustments.

To continuously optimize the learning outcomes and generate results that better fit the context, researchers need to increase the amount of training data for a language model in the partition for each language pair. Factors such as the complexity of the language model and learning algorithm used, as well as their error tolerance will further determine the total amount of data required for deep learning⁷. The quantity and quality of data will ultimately affect the outcome of its use on the algorithm model and are therefore crucial to the output quality of NMT.

Figure 5. SK hynix DDR5 DRAM CXL^TM Memory

Figure 6. SK hynix HBM3 DRAM

Therefore, NMT needs the support of large-scale floating-point operations to improve model performances such as inference speed, making the improved computation power an essential technological support⁸. To address processing demands like this, SK hynix recently developed the HBM3 DRAM and the first CXL memory, which provide more advantages and possibilities for AI and deep learning with innovative, groundbreaking technologies. The expandable DDR5 DRAM-based CXL memory allows for flexible memory expansion compared to current server market, where the memory capacity and performance are fixed once the server platform is adopted. With a total bandwidth of 360-480 GB/s, and a total capacity of 1.15 TB, the CXL memory product highly appeals to a variety of fields that require high-performance computing. Moreover, HBM3 of SK hynix, which can process up to 819GB per second and adopts a 16-channel architecture that runs at 6.4Gbps, is now under mass production, supporting the significant improvement of computing performances. With the upgraded computing power, more languages can be converted into data, achieving better accuracy in translation and communication, helping create a future world without language barrier through high-performance technologies.

The constant breakthrough in technology is pushing our imagination of the future beyond past frameworks and limitations. What was once “impossible” gradually becomes the reality today. With globalization connecting countries and communities around the world, the developments of different societies are no longer isolated by geographic or language barriers. Every leap in the evolution of technology is helping create a world where our vision for the greater common good for all mankind can be eventually achieved.

While embracing challenges, constantly innovating, and striving to offer better products and technologies, SK hynix keeps a close eye on the global community and hopes to make this world a better place with its expertise. Under the grandiose narrative of the world, tiny semiconductor chips hidden in devices are sparing no efforts to create infinite possibilities for our future society.

¹https://blog.busuu.com/most-spoken-languages-in-the-world/

²https://www.researchgate.net/figure/Barriers-for-effective-cross-cultural-communication-and-interaction-Source-made-by-the_fig2_338242321

³https://www.researchgate.net/figure/Barriers-for-effective-cross-cultural-communication-and-interaction-Source-made-by-the_fig2_338242321

⁴https://ai.facebook.com/research/no-language-left-behind/

⁵https://developer.nvidia.com/maxine

⁶https://www.forbes.com/sites/bernardmarr/2018/05/21/how-much-data-do-we-create-every-day-the-mind-blowing-stats-everyone-should-read/?sh=788642c860ba

⁷https://postindustria.com/how-much-data-is-required-for-machine-learning/

⁸https://www.sciencedirect.com/science/article/pii/S2666651020300024

The post Building Bridges with Machine Translation: Memory Chips Help Erase Language Barriers first appeared on SK hynix Newsroom.

When IBM-developed computer Watson beat out its human competitors on the quiz show Jeopardy in 2011, it was thought to be the beginning of the end of the superior reign of human intelligence. Watson brought discussions of AI to the mainstream. Its ability to apply machine learning to gather and analyze massive amounts of data in a flash was something most thought exclusive to sci-fi.

Quintillions of bytes of data are now being generated each day, with the amount of data generated by 2025 predicted to be 181 zettabytes. While this volume of data exceeds far beyond the realm of human consumption, cloud computing, faster processing, faster networks, and faster chips mean it can be processed and applied efficiently. AI isn’t a pipe dream – it’s a reality.

From Synapses to Circuits

Semiconductors supporting AI functions must capitalize on space and provide means for parallel processing for complex tasks. Enter, Processing in Memory chips. The so-called PIM chip integrates a processor with Random Access Memory (RAM) on a single memory module. This structure removes the boundary between memory and system semiconductors, allowing data storage and data processing to happen in the same place.

By eliminating the need for data to traverse modules, response times are greatly improved, allowing for real-time data processing. More traditional computer architectures, which manage processing and storage in separate modules, often fall prey to latency issues, commonly referred to as the von Neumann bottleneck. Adding processing functions to memory semiconductors presents a unique solution to overcome this long-standing problem.

SK hynix unveiled its next-generation PIM in February 2022 at ISSCC in San Francisco. The GDDR6-AiM (Accelerator in Memory) adds computational functions to GDDR6 memory chips, allowing for data to be processed at speeds of up to 16 Gbps.

GDDR6-AiM is also more energy efficient, reducing power consumption by 80% by removing data movement to the CPU and GPU. Advancing technology in a manner that supports a greener and more equitable world is an integral part of SK hynix future vision. GDDR6-AiM can help reduce carbon emissions and shrink the carbon footprint of any technology it’s applied to, advancing SK hynix’s ESG-related goals and expanding their positive impact across their clients’ industries.

While particularly effective in managing the needs of AI-based systems, PIM can be applied to a broad spectrum of technologies. Databases, query engines, data grids, and more all require some version of data storage and processing coupled with custom applications leveraging a variety of inputs.

The next generation of smart memory

Machine Learning vs. Deep Learning

Unbeknownst to many, artificial intelligence is a broad term that describes the science of creating machines that think like humans. The term machine learning marks functionalities that enable computers to perform tasks without explicit programming and includes deep learning, a subset that relies on artificial neural networks.

Deep learning can be seen as the most independent AI system as it manages both feature input and classification. These systems also require vast amounts of data and rely on parallel processes as their algorithms are primarily self-directed once trained.

AI machines, including deep learning models, are already a part of our lives. There are countless real-world AI applications, which only stand to increase. Everything from mobile devices to autonomous vehicles utilize AI models for tasks like location-based recommendation, auto-braking, camera-based object classification, and navigation through complex environments.

The art of computationally mimicking human intelligence takes many forms

Overcoming the Challenges

The road to PIM development was not without detours, roadblocks, and congestion. As the technology continues to advance, there are still obstacles to surmount across design, manufacturing, cost, and more.

Designing PIM requires the application of novel approaches to chip structures. Traditional semiconductors do not need to accommodate near-memory queues or perform parallel functions in the same way PIM chips do. Once onto the manufacturing stage, space and distance considerations become paramount. It is crucial to reduce how far signals must travel without increased cost or risk of thermal issues.

Furthermore, integrated chips such as PIM have an increased dependency on memory – a unique feature that is both a blessing and a curse. Any damage to the memory components could result in compromised data.

With the AI market expected to reach $190 billion by 2025, investment in AI is ripe. According to a Boston Consulting Group and MIT Sloan Management Review study, 83% of businesses say AI is a strategic priority. SK hynix will continue to advance its expertise in the area and lead this growing sector in the years to come.

The post Let PIM Do the Learning: The Brainpower Behind the AI Memory Chip first appeared on SK hynix Newsroom.

Most people try to block out annoying noises by buying headphones with better sound insulation or increasing the volume on their devices, but both options can also damage hearing. What else can you do? Active Noise Control (ANC) provides a solution. While ANC utilizes advanced technology, it has actually been around for a long time. It matches one sound wave by generating a second sound wave with the same frequency but opposite amplitude to achieve noise reduction or elimination. In the past, this technology was mainly used for specialized purposes, such as aviation systems, high-end vehicles, military equipment, or high-end professional headphones.

However, the continuous development of technology and the constant expansion of the consumer market have led to widespread adoption and adaptation of ANC technology for everyday life.

Everyday applications for ANC technology

ANC may be an unfamiliar concept for most people, but it is actually an important part of our everyday life. Headphones, cars, and many home appliances use ANC technology.

1. Earphones / Headphones 

Image Download

ANC is gaining more and more popularity among younger generations because its noise reduction technology drastically improves the user experience. Most people have different sensitivities to sound quality, so earphones and headphones ranging from 20 to 200 dollars often fail to demonstrate any improvement or real difference to consumers. However, ANC technology creates a clear difference in quality. In a noisy city, wearing ANC headphones is like having your own small world. Users are free to listen to songs, watch videos, or receive phone calls without interruptions or distractions.

ANC headphones employ a simple process to achieve noise reduction. First, the headphones receive an outside sound wave, and the sensors embedded in them convert the sound into an electromagnetic wave. Then the noise reduction system generates a sound wave of the opposite amplitude to reduce or even eliminate the noise.

There are currently a variety of ANC headphones on the market, and several famous brands, such as Sony and Bose, offer different models with ANC.

For instance, Sony offers a diverse range of unique headphones to meet consumer needs for a variety of scenarios. Bose headphones even redefine noise reduction with their advanced technology. Specifically, Bose Noise Cancelling Headphones 700 offer various options for noise cancellation with levels from 0-10. Level 0 enables users to deactivate ANC function to listen to both their content and ambient noises, such as people speaking or cars honking. Once users no longer need to listen to ambient noises, they can return to their own private world by switching to level 10 for complete noise cancellation.

2. Cars

Do cars need noise reduction technology? Of course. There are basically four sources of noise for cars: the engine, tires, air, and resonant vibrations from the car body. Many high-end cars combine cotton insulation, double-pane windows, and quiet tires with ANC technology to minimize noise.

Cadillac is a leader in applying ANC technology to cars. It prioritizes a quiet in-vehicle atmosphere as its most basic requirement for establishing cars as a “third space”, or a place of rest outside of home and work, during travel. Cadillac proved its commitment to this concept as early as 2013 with the launch of its XTS luxury sedan equipped with the premium Bose 5.1 surround sound system. The system collected noises through microphones or sensors in different parts of the vehicle, including the acceleration and vibration sounds from the vehicle body. It then actively played sound waves with opposite amplitude with its Digital Signal Processor (DSP) to cancel out the sound wave of the noise.

Figure 1. Active Noise Reduction in Automobiles¹

Image Download

The models following the Cadillac XTS, including the Cadillac CTS, CT6, and XT5, were also equipped with ANC technology. However, ANC is not limited to Cadillac, and use of ANC technology is now widespread among different car companies and models. Car models with high-performance ANC include Honda Accord, Ford Mondeo, Buick Envision, and more.

3. Homes

Image Download

Increasing living standards are driving a growing demand for indoor noise reduction. Ambient sounds outside of windows, the activities of neighbors, and the operation of home appliances can create a high level of noise pollution and adversely affect occupants. Purchasing appliances with ANC technology is currently one way to minimize such noise pollution. However, developing ANC for 3D open sound fields to reduce noise pollution will be the most effective solution for controlling indoor noise in the future.

Current ANC technology can manage noise reduction frequencies of about 50Hz-1,000Hz. The noise reduction range can also manage noises with medium and low frequencies, such as noises from tables, chairs, and footsteps in the upstairs room. This technology provides several advantages over traditional soundproof insulation. It advances technology and develops environmentally friendly practices by:

• Reducing the use of thick and heavy materials for less pollution.
• Simplifying installation, maintenance, and replacement.
• Increasing the efficiency of noise reduction.

Mainstream technologies for noise reduction

The three most essential components of ANC are ADC, DSP, and DAC. The Analog-to-Digital Converter (ADC) converts environmental sound waves into digital signals. Digital signal processor (DSP) analyzes the signals to generate opposite digital signals. Finally, the Digital-to-Analog Converter (DAC) converts the digital signals back into sound waves.

Figure 2. Flowchart depicting Digital ANC process²

Image Download

Because previous generations of DSP chips lacked the power to support complex computing in real time, analog ANC was the only option on the market. However, the rapid development of DSP chips has helped create three types of ANC: feedforward ANC, feedback ANC, and hybrid ANC.

Feedforward ANC places a single microphone on the outside of the earpiece. It receives an outside sound signal, transmits it to the ANC hardware, and cancels the noise before the user hears it. However, using a single microphone on the exterior increases exposure to background noise and makes achieving high sound quality difficult. As a result, the superior potential of hybrid ANC for noise reduction is attracting an increasing amount of attention. Hybrid ANC places a microphone on both the interior and the exterior of headphone earpieces to reduce sensitivity to background noise and suppress noises more effectively.

ADC, DSP, and DAC are all indispensable in ANC technology. The ADC converts the environmental noises into digital signals. The DSP uses advanced algorithms to reduce sound delay. Finally, the DAC creates an isolated sound environment in seconds to create an atmosphere of profound quiet.

Future trends for ANC semiconductors

In 2020, the substantial sales growth of the True Wireless Stereo (TWS) headphones helped drive the development of ANC technology. Many IC (integrated circuit) design firms have already made several breakthroughs in ANC technology. Design houses such as ADI, ams, and Unisoc are competing in the ANC market.

ANC semiconductors are adapting to the bigger market by becoming smaller and lowering power consumption. Many IC firms are prioritizing smart technology in developing methods to align ANC semiconductors more closely with technological evolution in the future. ANC products can employ smart technology with optimized noise reduction models and algorithms for greater adaptability to diverse application scenarios. These products will eliminate manual noise reduction configurations and reduce power consumption with smart technology.

The intellectualization of ANC semiconductor requires a lot of big data processing and machine learning, and it is known that semiconductor memory plays a vital role in the big data processing. It is believed that the ANC industry along with the rapid development of semiconductor memory can become more intelligent and being applied to more applications in the future. SK hynix’s DRAM and NAND flash memory products will contribute to the development of this industry as well.

Companies are also trying to advance integration across automotive, industrial, construction, and other sectors to incorporate semiconductor chips into a variety of processes and environmental improvement efforts. Their end goal is to build an effective system for noise reduction in a unified 3D environment.

The future possibilities for ANC technology are endless. It can be combined with cameras to intelligently monitor noise pollution from vehicles on the road. Cars, trains, and planes can use ANC technology to provide quiet and comfortable trips. Construction noise pollution can be dramatically reduced to benefit nearby residents, and construction sites can operate longer with increased productivity. Finally, ANC technology with smart features will redefine how we live by enabling us to enjoy the sounds of nature.

¹Reference: https://www.eefocus.com/automobile-electronics/463870?utm_source=newsletter&utm_medium=email&utm_campaign=mar_new_article
²Reference: https://www.edn.com/a-perspective-on-digital-anc-solutions-in-a-low-latency-dominated-world/

The post ANC technology: Your Everyday Life Evolution first appeared on SK hynix Newsroom.