Developing in React Native using Atom and Flow

| February 8, 2017

article image
One drawback of writing in React Native is the lack of type-checking that is common in Javascript. While using a separate language like TypeScript is a possibility, it also requires full commitment to the build process and for developers familiar with Javascript to learn a new technology. Flow is a popular alternative that actually ships with React Native. This post intends to show you how to get the flow errors and warnings in your Atom editor as well as how to set up pre-commit to restrict committing code that has Flow errors.

Spotlight

ADLINK Technology

ADLINK Technology is a global leader of Edge Computing with a mission to reduce the complexity of building IIoT systems. ADLINK provides hardware, connectivity, and software to form IIoT solutions for manufacturing, networking & communications, medical, transportation, oil & gas, and government & defense industries. Our solutions can include embedded building blocks and intelligent computing platforms, fully featured edge platforms, data connectivity and extraction devices, secure software for data movement, and micro services to monitor, manage, and analyze data-streaming assets and devices.

OTHER ARTICLES

Introduction To Artificial Intelligence And Machine Learning

Article | June 23, 2021

Lately, we all often come across two very hot buzzwords — Artificial Intelligence (AI) and Machine Learning (ML). Perhaps the impact of artificial intelligence and machine learning on today’s business world is more than our daily lives. According to a Bloomberg report, around $300 million were invested in 2014 to promote AI-powered startups. It was 300% more than the previous year’s investment in venture capital. It’s hard to deny the fact that artificial intelligence and machine learning are all around us. Whether it is about protecting confidential information at work or just playing your favourite games on PS5, AI and ML are there. Researchers, scientists, computer engineers, and analysts are working hard together to pass on human-like intelligence in machines so that they can think and act according to real-life scenarios. Businesses have changed their approach to AI keeping enterprise adoption in mind rather than treating it as just a research topic. Tech giants such as Google, Facebook, Microsoft have already invested billions in Artificial Intelligence and Machine Learning and already have started to reshape the customer experience. But the AI and ML incorporation we see today is just the tip of an iceberg. In the coming years, you will see them take over products and services one after another. What Is Artificial Intelligence and Machine Learning? It is nowadays common to see several companies marketing themselves as AI-powered startups even though their operations don’t really revolve around AI. To understand this type of gimmicky marketing, it is essential to first understand what Artificial Intelligence and Machine Learning are. Let’s be clear in the beginning about one fact — AI and ML are not the same things. If you think they are, kill this perception before it makes things very confusing. Both these terms crop up especially when the discussion is about the use of Artificial Intelligence in marketing, the use of Machine Learning in marketing, analytics, Big Data, and the modern-day tech that is transforming the world. To ease down the learning, here’s the best answer: Artificial Intelligence is a science used to develop systems that can mimic decision-making and behaviour like humans. In simple words, the main application of Artificial Intelligence is to make intelligent machines. Machine Learning is the subset of artificial intelligence that uses data to perform tasks. It involves designing and applying the data models or algorithms that can learn from their past experiences. There’s a subset of Machine Learning, too — Deep Learning. It counts on multilayered neural networks to perform tasks. Early Days of Artificial Intelligence The early mentions of AI trace back to Greek mythologies that have stories of a mechanical man that could mimic our own behaviour. Plus, the early computers were termed as “logical machines'' in Europe. These machines could solve arithmetic operations and even store memory. Scientists, fundamentally, were inspired by them to create mechanical brains. Over time, technology got more and more modern. And, our understanding of how the human mind works improved. Both these factors lead to the current AI revolution. Today, the use of AI is more focused on mimicking the decision-making process of humans rather than performing complex calculations. The prime motive of this is to allow machines to think and act more like humans. AI-powered machines that are designed to act intelligently come into two basic groups — General AI and Applied AI. General AIs are relatively less common and can theoretically handle any task. The most exciting improvements in the field of AI are happening in this specific area. In fact, it’s generalized AI that led to the rise of Machine Learning. On the other hand, applied AIs are designed to perform relatively smaller tasks like smartly trading shares and stocks, or guiding an autonomous vehicle to its destination, etc. The Rise of Machine Learning As mentioned earlier, Machine Learning is a subset of AI and can also be treated as the current state-of-the-art. It came into reality primarily because of the two major breakthroughs — the rise of the internet and human realization. In 1959, an American pioneer in the field of computer gaming and AI, Arthur Samual, realized that it can be possible to teach machines how to learn to perform tasks themselves rather than us telling them how to. As long as the emergence of the internet is concerned, that helped scientists with tons of digital information that could be analysed for the betterment of AI and eventually, ML. After these innovations, it was more efficient for scientists and engineers to program machines in a way that they learn to think like humans and then connect them to the internet so that they have all the needed information. Vertical AI And Horizontal AI No matter what kind of AI research it is, knowledge engineering is its essential part. Machines need plenty of information to think and act like humans. Therefore, AI needs access to objects, categories, properties, and relations between them to apply knowledge engineering. AI is responsible for generating analytical reasoning power, problem-solving abilities, and common sense in machines. And, it is not an easy task! The way AI serves us can be divided into two parts — Vertical AI and Horizontal AI. Vertical AI is used to perform single jobs such as automating repetitive tasks, scheduling meetings, etc. Vertical AI bots are so accurate in performing a single job that people often mistake them for human beings. Horizontal AI, on the other hand, can handle more than one task at the same time. The best examples of horizontal AI are Alexa, Siri, and Cortana. Different Types of Machine Learning ML can be best used to fix complex tasks such as enabling self-driving cars, face recognition, credit card fraud detection, etc. It uses huge, complex algorithms that keep on iterating frequently over big data sets. The following are the 3 major Machine Learning areas: ● Reinforcement Learning ● Unsupervised Learning ● Supervised Learning Reinforcement Learning In reinforcement machine learning, algorithms allow machines and software agents to automate ideal behaviour within a particular context to improve the performance of an overall system. It is characterised by learning problems rather than learning methods. If any method can solve a problem, it can be a reinforcement learning method. This Machine Learning technique assumes that the dynamic environment is connected to a software agent such as a computer program, bot, or robot. Ultimately, it chooses a specific action in order to rapidly deliver the most efficient result. Unsupervised Learning Due to the involvement of unclustered data, unsupervised machine learning is more complex than others. With it, the machine has to learn independently without any supervision. No fixed or correct solution is provided for any problem in this technique. The algorithm has to identify the data patterns and find the solution. The recommendation engines we see on several eCommerce websites and Facebook friend requests suggestions are the best examples of this sort of Machine Learning. Supervised Learning Training datasets are used in supervised learning. The algorithms are created in such a way that they can analyse the data patterns and develop an inferred function. The produced correct solution is then used to map new examples. The best example of supervised machine learning is credit card fraud detection. Final Words Artificial Intelligence and Machine Learning never fall short to surprise us with their exciting innovations. Their impact has reached all the industries including eCommerce, customer service, finance, education, healthcare, pharma, infrastructure security, and whatnot. Needless to say, all these industries are very keen on reaping all the benefits of Artificial Intelligence and Machine Learning. The human-like AI was an inevitable thing as most technologists thought. Today, we are indeed closer to this goal than ever. This exciting journey in the past couple of years is the result of how we predict AL and ML works. FAQs Why is AI Marketing important? With AI marketing, businesses and marketers can analyse and consolidate a large amount of data from emails, social media, and other platforms faster. The achieved insights can be used to improve campaign performance and eventually boost the returns on investment in a relatively lesser time. AI marketing is the best and the most efficient way to eliminate the risks of human errors while optimizing and streamlining the campaigns more effectively. The following benefits of AI marketing justify the attention it has received all over the world. ● A better understanding of your consumers ● Optimization of digital advertising campaigns ● Offer comprehensive customer profiles ● Allow real-time interactions with consumers ● Refined content delivery ● Reduced marketing costs ● Improved ROI Is artificial intelligence and machine learning the same? The straight answer to this question is NO. They are not the same thing. AI allows machines to learn human behaviour while ML is the subset of AI that teaches machines to learn on their own with the help of past data. Does AI need machine learning? Fundamentally, ML is not required for AI as AI systems do not need to be pre-programmed. Instead of such software agents, they get help from algorithms that can use their own intelligence to solve queries. These can be Machine Learning algorithms such as Deep Learning neural networks and Reinforcement Learning algorithms. { "@context": "https://schema.org", "@type": "FAQPage", "mainEntity": [{ "@type": "Question", "name": "Why is AI Marketing important?", "acceptedAnswer": { "@type": "Answer", "text": "With AI marketing , businesses and marketers can analyse and consolidate a large amount of data from emails, social media, and other platforms faster. The achieved insights can be used to improve campaign performance and eventually boost the returns on investment in a relatively lesser time. AI marketing is the best and the most efficient way to eliminate the risks of human errors while optimizing and streamlining the campaigns more effectively. The following benefits of AI marketing justify the attention it has received all over the world. A better understanding of your consumers Optimization of digital advertising campaigns Offer comprehensive customer profiles Allow real-time interactions with consumers Refined content delivery Reduced marketing costs Improved ROI" } },{ "@type": "Question", "name": "Is artificial intelligence and machine learning the same?", "acceptedAnswer": { "@type": "Answer", "text": "The straight answer to this question is NO. They are not the same thing. AI allows machines to learn human behaviour while ML is the subset of AI that teaches machines to learn on their own with the help of past data." } },{ "@type": "Question", "name": "Does AI need machine learning?", "acceptedAnswer": { "@type": "Answer", "text": "Fundamentally, ML is not required for AI as AI systems do not need to be pre-programmed. Instead of such software agents, they get help from algorithms that can use their own intelligence to solve queries. These can be Machine Learning algorithms such as Deep Learning neural networks and Reinforcement Learning algorithms." } }] }

Read More

How safe is our quantum future?

Article | June 28, 2021

If you're my age, you will remember the critical premise of the 1992 classic "Sneakers" premise, starring Robert Redford and Ben Kingsley - a top-secret black box that can break the encryption of any computer system. Quantum computing is that "black box." In the next 2-7 years, quantum computers could change the face of cybersecurity. Once they can factor products of large prime numbers (the basis of current cryptography) (expected between 2024 and 2030) – existing cyber-defense mechanisms will be rendered obsolete. We need to plan for encryption in the quantum future. What is Quantum Computing? Classical computers use binary arithmetic - all numbers are a sequence of bits - either a 1 or a 0. However, a quantum bit (qubit) exists not as a 0 or 1 but as a superposition of the two (think Schrödinger's cat). Every additional qubit doubles the processing power of a quantum computer, allowing it to execute multiple computational paths simultaneously. Similarly, as per Grover’s algorithm, it is a known fact that quantum computing divides the key space of symmetric cryptography algorithms by two, meaning that their key sizes have to be doubled to keep the safety margin of today. In October 2019, Google demonstrated quantum supremacy with Sycamore. It performed a series of operations in 200 seconds that Google claimed would take a supercomputer about 10,000 years to complete. In December 2020, physicists from the University of Science and Technology of China in Shanghai performed a Gaussian boson sampling technique with their photon-based quantum computer, named Jiuzhang. They declared that Sunway TaihuLight (the fourth fastest supercomputer in the world) would require 2.5 billion years (approx. half the age of the Earth) to finish the computations done by their quantum computer in a mere 200 seconds. Cryptography: The gatekeepers of security As the wise Spider-Man said – "With great power comes great responsibility. And great risk.” Much of the world's encrypted data is protected using mathematical equations with millions of reasonable solutions. These encryption models are too complicated for even supercomputers to solve within an acceptable period, which quantum systems can quickly solve. Modern cryptography relies on symmetric and asymmetric standards. The significant difference is that symmetric cryptography is based on substitution and permutation (there is no underlying mathematical assumption) and uses a single key for encryption and decryption. In contrast, asymmetric key / public key cryptography uses two different keys for encryption and decryption. Since the mid-90s, researchers have theorized that quantum computers can break current public-key cryptographic (PKC) systems. Their ability to concurrently test multiple hypotheses (using Shor's factorization OR Grover's exhaustive search) at unprecedented speeds will make both asymmetric and symmetric cryptosystems redundant. Understanding 5G 5G is one of the most eagerly awaited technologies in the digital world, and with good reason. In the years ahead, 5G coupled with IoT, could revolutionize the integration of digital and physical worlds. What sets it apart from its predecessor? 5G speed - it is nearly 20x faster than 4G. An average-length movie takes 6 minutes to download on 4G and less than 20 seconds on 5G. 5G supports 10x more devices per sq. km. It will seamlessly handle many more devices within the same area – a boost for IoT infrastructure. 5G latency is 25x less than 4G. According to McKinsey, 5G will speed up the mainstream adoption of IoT across multiple industries: Transport, Manufacturing, Healthcare, to name a few. 5G and Quantum – the Perfect Storm While quantum systems provide the compute, 5G provides the channel to connect more than just mobile networks (self-driving cars, personal medical tech), thus expanding the 'threat surface.' In a 5G world, secured communications are a critical component of connectivity, and post-quantum cryptography will play a key role. Researchers globally are devising ways to embed quantum-safe cryptography into 5G networks without compromising QoS. I even came across a patent for a quantum-resistant 5G SIM card by a Swiss company that set an industry best practice in ITU-T X.1811 for quantum-safe 5G. Cryptocurrency Wallets: A prime candidate for Quantum hacking Imagine you forget the password of your Bitcoin wallet, which in theory had millions of dollars in the balance. With a quantum computer, you could unlock your wallet and save yourself many worries, which worries all cryptographers. If malicious players had a quantum computer, the first thing they would try and break is the Elliptic Curve digital signature algorithm, reverse-engineer your private key, forge your digital signature, and subsequently empty your wallet. Thankfully, we are still years away from that scenario, yet that is a telling tale for designing national digital currencies that are supposed to withstand the test of time. Likewise, this vital subject – including applications with legal consequences such as smart contracts enabled by blockchain technologies, which share the same technical basis and, therefore, vulnerabilities to quantum IT -, would need a dedicated article, hopefully soon as time enables it! The real question is: when will quantum computers become a threat to public-key cryptography? As of December 2020, IBM claims to have a 65 qubit quantum computer and already delivering a 53-qubit model to a client (it would take around 1500 qubits to hack Bitcoin private keys). Quantum computers could achieve the required processing power range from as soon as 2024 to as far as 2040 per estimate. How do we solve it? Public Key Cryptography enables over 4.5 billion users to securely access over 200 million websites and engage in over $3 trillion of e-commerce transactions. Further, an estimated 20% of all IT applications rely on PKC and an even higher percentage on symmetric cryptography. According to Prof. Davor Pavuna of the École Polytechnique Fédérale de Lausanne, "several quantum prototypes might already become functional in 2023 (specifically in China)," and that potentially poses a severe protection challenge much earlier!" Many companies are developing "post-quantum cryptography" (PQC) or "quantum-safe cryptography" (QSC) – algorithms whose security is not degraded by any known quantum computing algorithms. Typical ones are McEliece cryptosystem, Lattice-based cryptosystems, Code-based Cryptography, and Hash-based cryptography. While these developments promise 'quantum resistance,' they only reflect our current knowledge of quantum computing capabilities and have a relatively low benchmark set for their security. These methods aim to create mathematical problems that are too difficult for even a quantum computer to solve, with the US National Institute of Standards and Technology (NIST) planning to recommend a PQC standard by 2022-23 and already having done so specifically for hash-based signatures. Similarly, German BSI issued official guidance for using post-quantum key exchange mechanisms, somewhat differing from NIST, and the IETF standardized two hash-based signature schemes, LMS and XMSS, independently, also with differences. Last but not least, the ITU-T issued without much publicity an amended recommendation on IPTV security X.1197 Amd1 that provides comprehensive guidance on state-of-the-art standard PQC options available as of late 2019, for use in multimedia transmission, with a corrigendum issued in early 2020. Applying the Solution Post Quantum cryptography is a developing field. Although these algorithms are quantum-resistant in theory, there is an unpredictability about their efficacy. Secondly, these algorithms are heavy on memory and compute requirements, making it challenging to apply them universally. On the other hand, symmetric cryptography is more efficient and shows more resilience to quantum IT, yet needs an upgrade to accommodate larger key sizes. One such system I came across was a patent of the aforementioned Swiss company is eAES®, which enhances AES’s quantum resistance. It makes safely increasing the key size a reality (as per NIST’s IR 8105 guidance), a claim confirmed in a report by their competitor Kudelski Security on the former’s implementation for Intel® processors. The transition to PQC standards requires a staged approach. To successfully navigate the impending cryptographic change, companies and governments must embrace crypto-agility - the ability to rapidly adapt and switch between multiple cryptographic standards at varying levels. We must support algorithms from different standardization bodies such as NIST, ETSI, the ITU-T, ISO/IEC, and the IEEE in a connected world with fractured standards. Building a global quantum security alliance We are just laying the foundations of this new security ecosystem; however, more work is needed to drive broader adoption. While the academic, innovation labs, and specialist technical communities are making some progress, cha

Read More

What is AI-guided selling?

Article | May 3, 2021

B2B selling has become increasingly complex for buyers and sellers. Buyers are inundated with content on a variety of channels, from an assortment of vendors. Sellers are often pulled in several directions, with many tasks and responsibilities to tackle. AI-guided selling is helping sellers navigate the complexity of digital-first sales cycles. Artificial intelligence (AI) and machine learning (ML) are helping sellers realize this vision by transforming data from content analytics into intelligent insights that enable go-to-market teams to make the most of every revenue moment. In this post, we’ll share the ins and outs of AI-guided selling, how it will affect go-to-market activities, and what it means for the future of sales and marketing.

Read More

MAKING ARTIFICIAL INTELLIGENCE SMARTER LIKE HUMAN BRAIN

Article | March 11, 2020

The latest advancements in Artificial Intelligence have been much tremendous and inspiring. It has become a part of everyday life for almost all consumers. In a large range of domains, the technology has transformed the way humans work and live. From smart home devices like Alexa, Siri, among others to large scale data security and fraud detection, all are inspired by and relied on AI. Despite this, there is still a large gap between current AI systems and human-like intelligence. Over time, the human brain has developed and advanced in order to respond to survival instincts, harness intellectual curiosity, and achieve demands of nature. While the human brain finds innovative ways to exceed its physical capabilities, human scientific pursuit amplified by the amalgamation of mathematics, algorithms, computational methods, and statistical models.

Read More

Spotlight

ADLINK Technology

ADLINK Technology is a global leader of Edge Computing with a mission to reduce the complexity of building IIoT systems. ADLINK provides hardware, connectivity, and software to form IIoT solutions for manufacturing, networking & communications, medical, transportation, oil & gas, and government & defense industries. Our solutions can include embedded building blocks and intelligent computing platforms, fully featured edge platforms, data connectivity and extraction devices, secure software for data movement, and micro services to monitor, manage, and analyze data-streaming assets and devices.

Events