DEV Community: Sachin Dev Duggal

Five Important Subsets of AI

Sachin Dev Duggal — Thu, 14 May 2020 06:45:33 +0000

As far as a simple definition, AI is the ability of a machine or a computer device to imitate human intelligence , secure from experiences, adapt to the most recent data and work people-like-exercises.

AI executes tasks intelligently that yield in creating enormous accuracy, flexibility, and productivity for the entire system. Tech chiefs are looking for some approaches to implement AI technologies into their organizations to draw obstruction and include values, for example, Artificial Intelligence is immovably utilized in the banking and media industry. There is a wide arrangement of methods that come in the space of AI, for example, linguistics, bias, vision, robotics, planning, natural language processing, decision science, etc. Let us learn about some of the major subfields of Artificial Intelligence in depth.

Neural Network

Joining cognitive science and machines to perform tasks, the neural network is a part of Engineer AI that utilizes nervous system science. Imitating the human mind where the human brain contains an unbounded number of neurons and to code brain-neurons into a system or a machine is the thing that the neural network functions.

Neural network and machine learning combinedly tackle numerous intricate tasks effortlessly while a large number of these tasks can be automated. NLTK is your sacred goal library that is utilized in NLP. Ace all the modules in it and you’ll be a professional text analyzer instantly. Other Python libraries include pandas, NumPy, text blob, matplotlib, wordcloud.

Deep Learning

An explainer article by AI software organization Pathmind offers a valuable analogy: Think of a lot of Russian dolls settled within one another. “Profound learning is a subset of machine learning and machine learning is a subset of Artificial Intelligence, which is an umbrella term for any computer program that accomplishes something smart.”

Deep learning utilizes alleged neural systems, which “learn from processing the labeled information provided during training and uses this answer key to realize what attributes of the information are expected to build the right yield,” as per one clarification given by deep Artificial Intelligence. “When an adequate number of models have been processed, the neural network can start to process new, inconspicuous sources of info and effectively return precise outcomes.”

Deep learning powers product and content recommendations for Amazon and Netflix. It works in the background of Google’s voice-and image-recognition algorithms. Its ability to break down a lot of high-dimensional information makes deep learning unmistakably appropriate for supercharging preventive maintenance frameworks.

Read full article @ https://www.analyticsinsight.net/five-important-subsets-of-artificial-intelligence/

SIGNAL for Covid-19

Sachin Dev Duggal — Sat, 02 May 2020 09:54:04 +0000

Abstract

In this position paper, we describe an approach to detecting and mitigating the spread of contagion across a large population. In light of the recent developments around the spread of the SARS-CoV-2 virus across multiple continents and geographies and the mortalities induced by the resultant disease, Covid-19, we believe that detecting, warning and arresting the spread of a contagion in a large population is one of society’s grand challenges. In this paper we present the blueprint of an end to end system that we believe is what is required to tackle such a societal challenge. This framework has been designed to operate at the level of an entire population such as a city, a state or even an entire nation. We believe this framework can equip decision makers, public health experts and governments with rich information and insights that can enable them to deploy the appropriate testing and vaccination schemes to eradicate the contagion from our societies.

Preamble

Early warning system for contagious disease spreading where testing is limited or logarithmic / not available and spread is exponential.
Assumption: Communication traces obtained by mobile phones are a good proxy of physical interaction network

Trade offs:

Coverage vs Sensitivity vs Specificity
Cost (prediction) vs error % (deaths)
Better to have more false positives than optimise costs (ie false negatives)
Invasiveness vs Effectiveness

Read full article https://medium.com/@aliamalhotra/sachin-dev-duggal-engineer-ai-signal-for-covid-19-8beb3b27b5c2

Idea: Engineer Ai Machines During Coronavirus Pandemic To Combat Covid-19

Sachin Dev Duggal — Wed, 25 Mar 2020 06:28:31 +0000

The novel Coronavirus (CoViD-19) pandemic has afflicted 245,916, and killed 10,048 people, all under a span of 3 odd months (as of this writing). Within this short time, countries (and their governments) have been tossed in a state of ‘damage control’ – for you see, no country in this world had seen this coming (USA, for example, had fired their entire Pandemic Response Team, in May of 2018).

Humans forget that history repeats itself – the gross negligence towards Ebola, Swine Flu, Cholera, and the other pandemics of the past, which too came uninvited, had become our greatest undoing. To the devil advocating that “Nobody could’ve seen this coming!” – Well, no human could; but if we were to engineer AI to do exactly that, well then maybe…

Malicious Artificial Intelligence In The 21st Century

The term AI (Artificial Intelligence) gets thrown around by Silicon Valley savants, like Google’s Ian Goodfellow, for instance – who was recently hired by Apple to Engineer AI machines, as the Director of Machine Learning, in the Special Learning Group. Ian Goodfellow piloted the Generative Adversarial Network (GAN), for his PhD thesis in 2014. Long story short, GAN is the basis for generating Deepfake videos; wherein a person’s facial likeness is replaced by someone else’s, to ‘doctor’ fake news, revenge porn, celebrity pornographic fakes, and hoaxes, among other such unimaginable malicious uses.

This is merely a microcosm of the potential that professionals have been able to engineer AI bots for, ever since the big Artificial Intelligence boom of the 2010’s. Ian Goodfellow might never have intended for his admittedly revolutionary software, to be used for malicious purposes – yet, here we are!

Voice mimicking AI are bleeding corporations worth hundreds of millions in wire fraud, as observed by Vijay Balasubramanian (CEO of Pindrop, an American security company dealing in tele-frauds), who said of the voice-deepfakes - “we’ve seen only a handful of cases, but the amount of money stolen can reach as high as $17 million individually”.

Besides a couple of these offshoot cases historically, AI has been overwhelmingly used for benevolence, rather than malevolence.

Coronavirus: Through The Lens Of AI

The oldest of adages, often referenced with respect to medicine and healthcare, is that “Prevention is better than cure” –a reminder that bears woeful repeating, belittling the very preaching it dispenses in the process. This, and the aforementioned, “those who forget history, are doomed to repeat it” – are constant reminders that humans tend to forget, but an AI never does.
From a bird’s eye view, AI is but a miniscule facet of the entire discipline. Machine Learning, is arguably the larger subset – A field of science established in 1955, when it was considered synonymous with robotics. However, nowadays, the conversation has shifted from humanoid cyborgs, to ethereal entities.

Machine Learning is the study of Algorithms and Statistical Models. A Computer is programmed to rely on patterns and interference, to build a mathematical model, rooted in “training data” (or sample data). Artificial Intelligence is the even smaller subset, which aims to simulate human behaviour into machine intelligence – scientists have been able engineer AI everywhere, from cybersecurity, and automobiles, to video games, and healthcare, among others. By 2025, nearly all spheres of industries will be AI-powered, in some shape or form.

This past Monday, the White House announced that they would partner with tech companies, and AI professionals, in an effort to fortify research on the coronavirus (Covid-19) outbreak. This will be achieved through algorithms, based in research, fuelled by an avalanche of data, from a generation of reference papers, and searching patterns - This is reparative, rather than preventive, however.
In another case, amateurs were able to engineer AI trackers for the coronavirus that drew in from reported cases of medical records – from within the media, newspapers, self-reports, and Facebook posts. Although this too was reparative, the call to action was faster than any government agency till that point (Microsoft engineered a similar AI tracker – nearly an entire month later!).

Coronavirus: A Measured, Human Response

The COrona VIrus Disease of 2019 (Covid-19) is an infectious disease, caused by the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) virus. Its escalation from a passing outbreak, to a pandemic, escalated within a matter of two months. For reference, The Black Death, which killed ~200 million people, manifested within a 7 year period, reaching current numbers by the 11th month. Humans, for all our advances as a society, simply aren’t equipped to be this prepared.

By March, 2020, the novel coronavirus had spread to nearly every major developed country in existence. Italy, being the hardest hit, lost more souls than China – a staggering 3,405 (and counting), to China’s 3,248. The coronavirus pandemic reached a >2% fatality rate, shaking the foundations of medical science, and putting their advancements into perspective. Modern medicine fell short, the governments fell short – moreover, humanity fell short!

Can You Engineer AI Bots To Combat A Pandemic?

Yes, absolutely! Although, the devil does lie in the details –

Earlier this week, UK officials upheld the decision to engineer AI, which would take the roles of those people, who had to miss work because of the Coronavirus pandemic. China proposed surveillance measures to track the spread of the contagion, by scanning crowded populations for citizens who were found coughing and sneezing. Scientists in Toronto were able to engineer AI that developed an anti-biotic, more powerful than what a human could.

Measures like these and more have come to light during the fallout of the coronavirus outbreak. These were latent until a catastrophe brought them to the limelight. Necessity is the mother of invention – another adage that hits the crises home, and really puts things into perspective.

That being said, Artificial Intelligence can not only be used for reparative measures, but for prophecy as well. On 30th December, 2019: an AI-powered algorithm called Bluedot, eerily detected the possibility of an outbreak out of Wuhan, China. The team engineering Bluedot, received a notification about an “unexplained phenomenon” coming out of Wuhan. The AI was quick to sound an alarm on this, as a possible “Health Risk”. This was largely ignored until the next few weeks, when the pandemic did start unfolding.

This goes on to show that, although humans lack the fortitude to track every possible case scenario under the sun, their machine assistants can actually do exactly that. Humans can definitely Engineer AI machines that will aid them to run a well-oiled, functioning operation. The AI will deliver always – but that is subject to what its human masters direct it towards.

Countries To Engineer AI For Futuristic Policing (Just Like In The Movies)

Sachin Dev Duggal — Mon, 23 Mar 2020 10:22:02 +0000

Surveillance and Artificial Intelligence go hand in hand. Scientists managed to engineer AI facial tracking bots that are able to narrowly track down persons of Interest, with a 12.5% accuracy. This might seem low now, but in the future, this number is bound to rise. Now, with the recent Coronavirus Pandemic looming over our heads, scientists at Carnegie Mellon, who developed AI-powered traffic stops in 2014, have been able engineer AI in CCTV cameras, which employ “patient tracking” software, used identify those potential infected at risk.

Policing Through AI

Recently, public sentiments against the police have soured. TV Shows like the 2019 version of The Watchmen, have shed the blue bloods in a dire light. It’s safe to say that the people working behind the uniform are people as well. Policing software has existed for quite some time now, although not as effective as one would like. With the recent Coronavirus (Covid-19) pandemic, a new facial recognition software has sprung up, that can partially do the police’s job for them.

Developed by China, it cannot track facial features, but only potentially criminal behavior. Before you have thoughts of dictatorial regimes popping through your head, criminal behavior here implies deluding or hiding from the authorities, and withholding cases of Covid-19. Couching incessantly, sneezing, looking pale, etc. are tells for this Engineer AI.

While we have a long way to implement proper police reform, steps like these would ease soft-criminality, by allowing officials to engineer AI tracking machines, who look out for the Coronavirus now, and will criminals later.

5 Trends In Cloud Computing To Look Out For in 2020

Sachin Dev Duggal — Sat, 21 Mar 2020 10:14:00 +0000

Cloud Computing is more than a buzzword – it’s the great leveller! Irrespective of how big, or small the enterprise, Cloud based services are ubiquitously enjoyed by everyone. Weather you sort logistics, or Engineer AI – Cloud Computing is a service available to everyone (and for good reason).

So what is Cloud Computing? Simply put, Cloud Computing lets anybody using a network of remote servers, to host on the internet, rather than to store, process, and manage data, on a personal computer. This is how companies like Microsoft, Amazon, Google, and IBM are looking to make their billions in the coming years.

The growth of Cloud Computing is exponential – in the words of Sachin Dev Duggal, CEO of Engineer.ai(also known as Builder.ai), a cloud based enterprise, “cloud based services will enhance currently existing industries... such that 9% of the young Indian workforce will be employed in jobs by 2022 (under cloud computing), which do not even exist today”.

The Cloud based market is expected to grow to a $214 billion dollar industry in 2019 (up from $182.4 billion in 2018). It begs to understand this rapidly burgeoning industry at hand, and look at the 5 emerging trends in Cloud Computing.

Quantum Computing

When scientists started to engineer AI bots, as nothing more than calculating devices, the scepticism was rife over its application – now, Artificial Intelligence is more than just a passing fad, but a way of life.

Quantum computing accrues the same amount of scepticism nowadays. Quantum computers use principles based in quantum mechanics, such as superposition (where a quantum entity is able to exist in multiple states at the same time) to power its hardware.

Quantum computers are remarkably faster than the antiquated PC’s we use now. Such is faster pace of work is complemented seamlessly by cloud computing. Going forward, these two terms would be inexplicably linked with each other forever.

Public Cloud Services

Cloud based services are not exclusive to corporations who merely “sort logistics and engineer AI bots” – the standard ‘layman’ is on to it as well (without being aware of it). General public usage of cloud services is growing 3 times faster than it is for corporates. Basic communication tools like Whatsapp (which is used by nearly 1.5 billion people), uses cloud based services inadvertently.

Combine that with other platforms like Facebook, and Google – cloud computing commands a lion’s share for general public, than otherwise. Jeff Bezos, who’s AWS (Amazon Web Services) happens to be the leading cloud based service currently, said of Public Cloud that it would be “hard to overstate how big of an impact it’s going to have on society over the next 20 years”.

Hybrid Cloud Computing

There are two facets for the emergence of a Hybrid Cloud – an on-premises private cloud, and a third-party, public cloud. This is crucial for spheres, such as education, and app development. The edu-tech market, according to Sachin Duggal, is achieving a Compound Annual Growth Rate of 49% - a Hybrid Cloud will essentially eliminate the need for classrooms and traditional educational tools.
Hybrid Cloud architectures could help better integrate on-premises and cloud operations, for a broader use, in cases of setting cloud services, tools, API’s, and on-premises enterprises.

Paas, Saas, and IaaS

PaaS (Platform as a Service), SaaS (Software as a Service), and IaaS (Infrastructure as a Service) – are virtually contingent to cloud computing, without being intrusive about it. Ubiquitous applications like Dropbox (SaaS), Windows Azure (PaaS), and Amazon Web Services (IaaS), go hand in hand with cloud based services.

The rise of these integrated applications empowers cloud services and infrastructure, to achieve a tonne of previously unimaginable jobs (for example, but not limited to sorting logistics and engineering AI). Between PaaS, SaaS, and IaaS, the cloud based service net will add nearly 125 billion US Dollars to software vendors by 2020.

Global-Mobile Cloud

Of the companies that do not utilize cloud computing right now, 17% will, by the end of 2020. Moreover, this has become a global phenomenon. More and more people, corporations, applications and governments are accessing cloud based services increasingly daily.

This, along with the rise of handheld devices, is setting a precedent for a world, where everything would be linked to every other device – on the go! Seamless interaction between entities will become an increasing reality, no matter where anyone is.

This could bolster the case for multi-currency, multilingual databases, to be accessed almost instantly – eliminating any sense of obstacle and cultural barrier thereof.