Watch Out Microsoft Warns Android Users About A New Ransomware!

Watch Out Microsoft Warns Android Users About A New Ransomware!

Microsoft has warned about a new strain of mobile ransomware that takes advantage of incoming call notifications and Android’s Home button to lock the device behind a ransom note.

The findings concern a variant of a known Android ransomware family dubbed “MalLocker.B” which has now resurfaced with new techniques, including a novel means to deliver the ransom demand on infected devices as well as an obfuscation mechanism to evade security solutions.

The development comes amid a huge surge in ransomware attacks against critical infrastructure across sectors, with a 50% increase in the daily average of ransomware attacks in the last three months compared to the first half of the year, and cybercriminals increasingly incorporating double extortion in their playbook.

MalLocker has been known for being hosted on malicious websites and circulated on online forums using various social engineering lures by masquerading as popular apps, cracked games, or video players.

Previous instances of Android ransomware have exploited Android accessibility features or permission called “SYSTEM_ALERT_WINDOW” to display a persistent window atop all other screens to display the ransom note, which typically masquerade as fake police notices or alerts about purportedly finding explicit images on the device.

But just as anti-malware software began detecting this behavior, the new Android ransomware variant has evolved its strategy to overcome this barrier. What’s changed with MalLocker.B is the method by which it achieves the same goal via an entirely new tactic.


To do so, it leverages the “call” notification that’s used to alert the user about incoming calls in order to display a window that covers the entire area of the screen, and subsequently combines it with a Home or Recents keypress to trigger the ransom note to the foreground and prevent the victim from switching to any other screen.

“This creates a chain of events that triggers the automatic pop-up of the ransomware screen without doing infinite redraw or posing as a system window,” Microsoft said.

Aside from incrementally building on an array of aforementioned techniques to show the ransomware screen, the company also noted the presence of a yet-to-be-integrated machine learning model that could be used to fit the ransom note image within the screen without distortion, hinting at the next stage evolution of the malware.

Furthermore, in an attempt to mask its true purpose, the ransomware code is heavily obfuscated and made unreadable through name mangling and deliberate use of meaningless variable names and junk code to thwart analysis, the company said.

“This new mobile ransomware variant is an important discovery because the malware exhibits behaviors that have not been seen before and could open doors for other malware to follow,” Microsoft 365 Defender Research Team said.

“It reinforces the need for comprehensive defense powered by broad visibility into attack surfaces as well as domain experts who track the threat landscape and uncover notable threats that might be hiding amidst massive threat data and signals.”

Original article by Ravie Lakshmanan and can be found here

 

Wolfe Systems is a Business Technology Solution provider. Our offerings are tailored to our clients requirements, we believe Technology is a tool that should be harnessed to assist businesses to achieve their goals. Find out how we can help you today. Contact us now!

Follow us on the WOLFE SYSTEMS socials for up to date tech trends,  information and cool facts

15 Top Social Media Sites & Platforms for 2020

15 Top Social Media Sites & Platforms for 2020

15 Top Social Media Sites & Platforms (A Handy 2020 Guide)

Wondering which social media sites your business, brand, or blog should be on in 2020?

You’ve come to the right place.

Because social media platforms are always jockeying for dominance, it can be surprisingly difficult to find articles with the most up-to-date information on Google.

The good news is we’ve done the legwork for you. Below, you’ll find the freshest data from sites like Statista to help you get the lowdown on the top social media platforms around the world.

We’ll look at cool features for each social network, go over who should (and should not) use them, and provide insights that’ll help you determine which ones are best for your business, brand, or blog.

Let’s dig in.

  1. Facebook

Facebook is the largest social media platform in the world with most of their daily users living outside of the USA

With people checking the site multiple times daily, sharing posts composed of text, photos, videos, and GIFs that other users comment on, share, and react to.

Just about every audience is on Facebook, and it is the most popular social media network for seniors — a rapidly growing demographic on the site.

A great platform to get people to know you. Know your brand and build a level of trust. And if you’re a content creator (whether it’s a blog post, infographic, or an article for a client), Facebook is one of the best social media platforms to share content.

Standout Features & Functionality

  • Audience Insights gives small businesses the power to leverage Facebook’s massive pool of aggregated data to help them reach their ideal audience.
  • Facebook 360 allows businesses to upload panoramic style photos from a smartphone, creating immersive, interactive experiences for their followers.
  • Facebook Live gives influencers an interactive medium for connecting with their audience.

 

  1. YouTube

When it comes to social networks, YouTube dominates video content sharing.

This massive video-sharing platform lets users post, comment on, and upvote or downvote music videos, TV shows, vlogs, educational content, live streams, movie trailers, and more.

They can also subscribe to other users’ channels and add videos to playlists.

YouTube functions as a massive search engine (2nd only to Google), making SEO an important consideration for content creators. Tons of businesses have leveraged this top social media site successfully

An astounding 500+ hours of video is uploaded to YouTube every minute. If you’re not committed to producing high-quality content, you won’t gain traction.

Standout Features

  • YouTube Cards visually enhance links to other videos and playlists, making your content more interactive. They can appear at any specified time during your video.
  • YouTube Live allows you to stand out by creating an engaging experience with your audience.
  • After reaching 1,000 followers, channels can post text, images, GIFs, live videos, and more within the Community tab to further connect with their audience

 

  1. WhatsApp

In a crowded messaging app market (QQ, Telegram, Snapchat, etc.) WhatsApp stands as the most popular, with more monthly active users than Facebook Messenger (both are owned by Facebook).

Users can send text and voice messages or make voice and video calls for free with this mobile app. However, unlike the app Viber, users cannot call non-app users’ numbers.

WhatsApp Business is a standalone app available on desktop, and many business owners use it as a handy customer service solution.

Standout Features

  • Free international text, voice, and video messaging (video chat with up to 4 people).
  • WhatsApp Business provides messaging tools for connecting with customers, including automated greeting and away messages, the ability to reuse messages for quick replies, labels for organizing contacts and chats, catalogs to show off your products and services, and a profile page with your basic business info.
  • Document sharing allows for easily sharing spreadsheets, slideshows, and other documents up to 100 MB (also free!).

 

  1. Facebook Messenger

Messenger, originally Facebook Chat, is a standalone messaging app and platform.

No Facebook account is necessary to use this social networking site to communicate via text messages, voice, or video chat.

More than an instant messaging app, users can also share photos, videos, stickers, and other file formats.

Messenger was recently redesigned to be lightweight and fast — deprioritizing a number of features, including chat bots that had become integral to many business’s customer communication strategy.

For a time, chatbots were a big marketing strategy on Messenger. Those days are over, with a shift toward user satisfaction. Respect your customers’ inbox!

Standout Features

  • US users can send and receive money through Messenger by connecting it to their debit card or PayPal account.
  • Location sharing (remember Foursquare and its check-ins?) makes it easy for your friends or customers to find you.
  • Messenger Rooms is a video chat feature similar to Zoom that allows up to 50 participants at a time.
  • Business solutions, like purchase tracking, notifications, and connecting customers with your customer service representatives make it easy to serve your customers.

 

  1. WeChat

WeChat (Weixin) is one of the most popular social media sites to come out of China.

Owned by Tencent, the parent company of QQ and QZone, this app does a little bit of everything.

It’s a messaging, social media, and mobile payment app rolled into one, and users also play games, shop, and access government services through the platform.

Like other social media apps, users can also share photos, make video calls, and text.

Advertisers on WeChat benefit from Facebook-esque customer data collection.

If your audience is in or from China — including tourists, students, and expats — then this all-in-one social media service is worth a look!

Standout Features

  • WeChat Pay allows users to shop and make money transfers.
  • Automated replies create a smooth customer service experience.
  • A customer history chat log is available for easy reference when corresponding with clients.
  • WeChat Shop platforms allow businesses to set up online stores that leverage features like WeChat Pay, phone number and address collection, retargeting, and customer service capabilities.

 

  1. Instagram

Instagram is a video and photo sharing social platform where users upload photos and short videos, often adding filters and other effects before sharing them with family and friends.

It’s owned by Facebook, which provides robust marketing data for reaching your audience.

Instagram is the perfect place to show off products and tell visually centered stories about your business. Whether that’s e-commerce or personal training, you’ll find that users on this platform are ready to buy.

A variety of brands thrive on Instagram, especially those in travel, beauty, and fashion — the largest demographic is under 35, urban females with above average income.

Instagram is growing every day. With increasingly sophisticated and powerful tools for marketers, it’s a great platform for growing your business.

Standout Features

  • Shoppable Tags allow users to click on products in images and go directly to a product page to make purchases.
  • Instagram Stories are short user-generated videos with various added enhancements and effects that disappear after 24 hours.
  • Instagram Live is a great way to engage your audience and interact with them in real time.

 

  1. TikTok

TikTok is a massively popular social network coming out of China that bills itself as the “leading destination for short-from mobile video.”

Users create various types of short (up to 15 seconds) looping videos to share, including comedy, talent, lip synching, dancing, stunts, and more.

Like other social networking platforms, its algorithm will learn what you like and show you similar content over time.

TikTok is a fun space to show your brand’s human side. If your audience is on this platform, it’s a great place to cultivate awareness.

Standout Features

  • In-app video editing allows users to customize background music, video effects, speed, filters, sounds, and more.
  • With the React feature, users can film their reactions to videos, showing up in a small window alongside the original content.
  • The Duet feature allows users to film themselves trying to sing the same song, perform the same dance, or do a parody as a joke alongside a piece of original content.

 

  1. QQ

QQ is another Chinese instant messaging platform.

Users can participate in group chats and send texts and audio messages — comparable to Facebook Messenger.

Although it’s primarily a desktop app, a lightweight and less functional mobile app is available.

If you’re doing business in China, then look into QQ further to see if it’s a good fit for your brand. Here is a more in-depth article on the platform.

Standout Features

  • A live translation feature for up to 50 languages makes it possible to connect internationally.
  • QQ Coin is a virtual currency used to purchase items for the user’s avatar or blog. Coins are accepted by some vendors for real goods.
  • Users can transfer files for free, with no limit on file size.

 

  1. Sina Weibo

Sina Weibo, simply called Weibo in China, is a microblogging social network.

Similar to Twitter, you can find journalists, celebrities and other public figures on the platform.

In addition to publishing text based updates (up to 2,000 characters), users can share music, videos, and images. They can also comment, follow, private message, search, and use @Usernames to tag others.

And, like Instagram, users can post images (9 max per post) and create Stories.

Standout Features

  • Weibo Polls show up in users’ timelines like a regular post, and are a great way to engage with and learn about your audience.
  • Businesses can claim their own hashtag, which can be used to increase brand awareness.
  • Weibo Fit tracks calories, walking distance, and other health data.

 

  1. QZone

With QZone, a bonded service to QQ, users share photos, watch videos, listen to music, keep diaries, play online games, shop, date, and blog.

Think of it like a mix of a social networking site and blogging platform. It’s similar to Facebook, and users can like, share, comment on posts.

They can also customize their profile backgrounds with accessories (mostly paid) and a song that plays in the background (like Myspace).

Standout Features

  • Brands can create customized “microsites” and other interactive marketing applications that function within the platform.
  • Users can have up to 1,000 QZone Albums, each with up to 10,000 photos.
  • A QZone Certified Space gives businesses extra features to help them stay connected with their audience.

 

  1. Reddit

The “front page of the internet,” Reddit is a forum where users can participate in thousands of communities called “subreddits,” covering any topic imaginable (and many more).

Content is user-generated and includes written posts, discussions, photos, videos, and links to articles.

Users often pose questions to the community (similar to Quora), or just browse for random stories like you would see on StumbleUpon back in the day.

If you want to gain insights into an audience’s thought process and communication style, then follow any subreddits they hang out in.

Reddit is an awesome resource for learning about topics and people. To learn more about marketing on Reddit, check out this article.

Standout Features

  • Ask Me Anything (AMAs) allow users to have live conversations with well known public figures like politicians, authors, actors, or people with interesting jobs.
  • Activities on reddit generate Karma, a point system that shows how much you participate and generate goodwill from other users.

 

  1. Kuaishou

Kuaishou, called Kwai outside of China, is a short-video sharing and live-streaming app.

Users share clips of stunts and pranks and live streams of video game play.

Kuaishou’s algorithm is designed for inclusivity, ensuring more visibility for a wider range of content.

This has opened the way for people from China’s rural regions to share their lifeways while generating extra income, which they do by promoting their businesses or linking out to ecommerce platforms.

E-commerce businesses (conversions are higher on this platform) selling to the Gen Z demographic in China’s smaller cities and rural regions.

Kuaishou’s algorithm is designed to be more equitable in providing exposure. That could be an opportunity to “break in” and be seen.

Standout Features

  • Kwai’s simple, easy to use interface allows anyone to quickly learn and use the app.
  • Live-streaming allows users to provide a variety of entertaining and educational experience.
  • Users can link out to e-commerce platforms.

 

  1. Snapchat

Snapchat is “a camera and messaging app that connects people with their friends,” specializing in ephemeral content — media that is only accessible for a short time (kind of like Tumblr’s ephemeral chat rooms).

Users share image and video massages called “snaps” can be modified with stickers, text, filters, and other effects. Snapchat users are always looking for interesting content. This is a great opportunity to tap into your creative side and engage younger audiences.

There are also private and public stories that show the last 24 hours of a user’s content.

Standout Features

  • Creators can add links to snaps, allowing them to send viewers to external websites.
  • Businesses can create a Sponsored Geo filter, a location specific image overlay that is only available at a specific physical location that “proves” the user was there.
  • The “Swipe Up to Call” and “Swipe Up to Text” features make it even easier for your audience to contact your business.

 

  1. Pinterest

People visit Pinterest for inspiration and to find and share new ideas.

This “productivity tool for planning your dreams” has virtual boards that users (called Pinners) fill with image-based “pins,” bookmarking them for future viewing.

Pins typically link to an external website and can be repinned from one user’s board to another.

Pinners view pins from all the other users and topics they’re following on their home feed.

If your audience is made up of women, and especially Millennial moms, then Pinterest could be perfect for you.

Pinterest users have proven to be ready to buy. The great thing is that Pinterest gives marketers plenty of tools to channel that buying intent.

Standout Features

  • “Lens” allows users to perform a visual search for objects they’ve captured on their phones.
  • “Catalogs” allows brands to upload their catalog into Pinterest and turn products into Product Pins.
  • “Rich Pins” come in 4 formats: Product Pins, Recipe pins, Article pins, App pins. Each with its own dynamic function.
  • “Shop the Look” pins allow Pinners to click on items and go to a product page to purchase.

 

  1. Twitter

Twitter is a microblogging platform where users primarily share short posts called tweets (280 character maximum) with their followers.

Users can then like, comment on, or retweet these posts to their followers. It’s like having a massive conversation with people all over the world.

Twitter is full of businesses and brands communicating with their audiences, and “40% of Twitter users reported purchasing something after seeing it on Twitter.”

Twitter is a great place to have a conversation with your audience. Show up and be ready to have a conversation with your people.

Standout Features

  • Users can see and subscribe to the curated feeds, called “Lists” of people they follow.
  • Twitter Polls allow users to ask their followers a question with 4 answers to choose from. This is an easy way to engage your audience and get feedback on ideas.
  • Hashtags were started on Twitter, and you can create branded hashtags to generate brand awareness and drive engagement.

Social Media Sites for Every Type of Business

You don’t need 15 social media accounts to get in front of your audience. In fact, that would be a terrible idea.

You’re better off focusing on one or two platforms where your target audience is most engaged. Each of the social networks we covered has its pros and cons. Pick one or two that you think could be a good fit for your brand, download their app from the Apple or Android store, and follow our links to learn more.

Now is a great time to start showing up in front of your audience. Study them, engage, add value, and before you know it they’ll become loyal fans and customers.

Full article by  James Everett Youngblood  can be found here

Wolfe Systems is a Business Technology Solution provider. Our offerings are tailored to our clients requirements, we believe Technology is a tool that should be harnessed to assist businesses to achieve their goals. Find out how we can help you today. Contact us now!

Follow us on the WOLFE SYSTEMS socials for up to date tech trends,  information and cool facts

 

A Student In Queensland Just Proved Paradox-Free Time Travel Is Possible

A Student In Queensland Just Proved Paradox-Free Time Travel Is Possible

A Student Just Proved Paradox-Free Time Travel Is Possible

watch symbolizing time and space

ARTPARTNER-IMAGESGETTY IMAGES

In a new peer-reviewed paper, a senior honors undergraduate says he has mathematically proven the physical feasibility of a specific kind of time travel. The paper appears in Classical and Quantum Gravity.

University of Queensland student Germain Tobar, who the university’s press release calls “prodigious,” worked with UQ physics professor Fabio Costa on this paper. In “Reversible dynamics with closed time-like curves and freedom of choice,” Tobar and Costa say they’ve found a middle ground in mathematics that solves a major logical paradox in one model of time travel. Let’s dig in.

The math itself is complex, but it boils down to something fairly simple. Time travel discussion focuses on closed time-like curves (CTCs), something Albert Einstein first posited. And Tobar and Costa say that as long as just two pieces of an entire scenario within a CTC are still in “causal order” when you leave, the rest is subject to local free will.

“Our results show that CTCs are not only compatible with determinism and with the local ‘free choice’ of operations, but also with a rich and diverse range of scenarios and dynamical processes,” their paper concludes.

In a university statement, Costa illustrates the science with an analogy:

“Say you travelled in time, in an attempt to stop COVID-19’s patient zero from being exposed to the virus. However if you stopped that individual from becoming infected, that would eliminate the motivation for you to go back and stop the pandemic in the first place. This is a paradox, an inconsistency that often leads people to think that time travel cannot occur in our universe. Logically it’s hard to accept because that would affect our freedom to make any arbitrary action. It would mean you can time travel, but you cannot do anything that would cause a paradox to occur.”

Some outcomes of this are grouped as the “butterfly effect,” which refers to unintended large consequences of small actions. But the real truth, in terms of the mathematical outcomes, is more like another classic parable: the monkey’s paw. Be careful what you wish for, and be careful what you time travel for. Tobar explains in the statement:

“In the coronavirus patient zero example, you might try and stop patient zero from becoming infected, but in doing so you would catch the virus and become patient zero, or someone else would. No matter what you did, the salient events would just recalibrate around you. Try as you might to create a paradox, the events will always adjust themselves, to avoid any inconsistency.”

While that sounds frustrating for the person trying to prevent a pandemic or kill Hitler, for mathematicians, it helps to smooth a fundamental speed bump in the way we think about time. It also fits with recent quantum findings from Los Alamos, for example, and the way random walk mathematics behave in one and two dimensions.

At the very least, this research suggests that anyone eventually designing a way to meaningfully travel in time could do so and experiment without an underlying fear of ruining the world—at least not right away.

Original article by

Germain Tobar’s paper can be found here

 

Wolfe Systems is a Business Technology Solution provider. Our offerings are tailored to our clients requirements, we believe Technology is a tool that should be harnessed to assist businesses to achieve their goals. Find out how we can help you today. Contact us now!

Follow us on the WOLFE SYSTEMS socials for up to date tech trends,  information and cool facts

Tiny Machine Learning: The Next AI Revolution

Tiny Machine Learning: The Next AI Revolution

The bigger model is not always the better model

 

Miniaturization of electronics started by NASA’s push became an entire consumer products industry. Now we’re carrying the complete works of Beethoven on a lapel pin listening to it in headphones. — Neil deGrasse Tyson, astrophysicist and science commentator

[…] the pervasiveness of ultra-low-power embedded devices, coupled with the introduction of embedded machine learning frameworks like TensorFlow Lite for Microcontrollers will enable the mass proliferation of AI-powered IoT devices. — Vijay Janapa Reddi, Associate Professor at Harvard University

Image for post

Overview of tiny machine learning (TinyML) with embedded devices.

This is the first in a series of articles on tiny machine learning. The goal of this article is to introduce the reader to the idea of tiny machine learning and its future potential. In-depth discussion of specific applications, implementations, and tutorials will follow in subsequent articles in the series.

Introduction

Over the past decade, we have witnessed the size of machine learning algorithms grow exponentially due to improvements in processor speeds and the advent of big data. Initially, models were small enough to run on local machines using one or more cores within the central processing unit (CPU).

Shortly after, computation using graphics processing units (GPUs) became necessary to handle larger datasets and became more readily available due to introduction of cloud-based services such as SaaS platforms (e.g., Google Colaboratory) and IaaS (e.g., Amazon EC2 Instances). At this time, algorithms could still be run on single machines.

More recently, we have seen the development of specialized application-specific integrated circuits (ASICs) and tensor processing units (TPUs), which can pack the power of ~8 GPUs. These devices have been augmented with the ability to distribute learning across multiple systems in an attempt to grow larger and larger models.

This came to a head recently with the release of the GPT-3 algorithm (released in May 2020), boasting a network architecture containing a staggering 175 billion neurons — more than double the number present in the human brain (~85 billion). This is more than 10x the number of neurons than the next-largest neural network ever created, Turing-NLG (released in February 2020, containing ~17.5 billion parameters). Some estimates claim that the model cost around $10 million dollars to train and used approximately 3 GWh of electricity (approximately the output of three nuclear power plants for an hour).

While the achievements of GPT-3 and Turing-NLG are laudable, naturally, this has led to some in the industry to criticize the increasingly large carbon footprint of the AI industry. However, it has also helped to stimulate interest within the AI community towards more energy-efficient computing. Such ideas, like more efficient algorithms, data representations, and computation have been the focus of a seemingly unrelated field for several years: tiny machine learning.

Tiny machine learning (tinyML) is the intersection of machine learning and embedded internet of things (IoT) devices. The field is an emerging engineering discipline that has the potential to revolutionize many industries.

The main industry beneficiaries of tinyML are in edge computing and energy-efficient computing. TinyML emerged from the concept of the internet of things (IoT). The traditional idea of IoT was that data would be sent from a local device to the cloud for processing. Some individuals raised certain concerns with this concept: privacy, latency, storage, and energy efficiency to name a few.

Energy Efficiency. Transmitting data (via wires or wirelessly) is very energy-intensive, around an order of magnitude more energy-intensive than onboard computations (specifically, multiply-accumulate units). Developing IoT systems that can perform their own data processing is the most energy-efficient method. AI pioneers have discussed this idea of “data-centric” computing (as opposed to the cloud model’s “compute-centric”) for some time and we are now beginning to see it play out.

Privacy. Transmitting data opens the potential for privacy violations. Such data could be intercepted by a malicious actor and becomes inherently less secure when warehoused in a singular location (such as the cloud). By keeping data primarily on the device and minimizing communications, this improves security and privacy.

Storage. For many IoT devices, the data they are obtaining is of no merit. Imagine a security camera recording the entrance to a building for 24 hours a day. For a large portion of the day, the camera footage is of no utility, because nothing is happening. By having a more intelligent system that only activates when necessary, lower storage capacity is necessary, and the amount of data necessary to transmit to the cloud is reduced.

Latency. For standard IoT devices, such as Amazon Alexa, these devices transmit data to the cloud for processing and then return a response based on the algorithm’s output. In this sense, the device is just a convenient gateway to a cloud model, like a carrier pigeon between yourself and Amazon’s servers. The device is pretty dumb and fully dependent on the speed of the internet to produce a result. If you have slow internet, Amazon Alexa will also become slow. For an intelligent IoT device with onboard automatic speech recognition, the latency is reduced because there is reduced (if not no) dependence on external communications.

These issues led to the development of edge computing, the idea of performing processing activities onboard of edge devices (devices at the “edge” of the cloud). These devices are highly resource-constrained in terms of memory, computation, and power, leading to the development of more efficient algorithms, data structures, and computational methods.

Such improvements are also applicable to larger models, which may lead to efficiency increases in machine learning models by orders of magnitude with no impact on model accuracy. As an example, the Bonsai algorithm developed by Microsoft can be as small as 2 KB but can have even better performance than a typical 40 MB kNN algorithm, or a 4 MB neural network. This result may not sound important, but the same accuracy on a model 1/10,000th of the size is quite impressive. A model this small can be run on an Arduino Uno, which has 2 KB RAM available — in short, you can now build such a machine learning model on a $5 microcontroller.

We are at an interesting crossroads where machine learning is bifurcating between two computing paradigms: compute-centric computing and data-centric computing. In the compute-centric paradigm, data is stockpiled and analyzed by instances in data centers, while in the data-centric paradigm, the processing is done locally at the origin of the data. Although we appear to be quickly moving towards a ceiling in the compute-centric paradigm, work in the data-centric paradigm has only just begun.

IoT devices and embedded machine learning models are becoming increasingly ubiquitous in the modern world (predicted more than 20 billion active devices by the end of 2020). Many of these you may not even have noticed. Smart doorbells, smart thermostats, a smartphone that “wakes up” when you say a couple of words, or even just pick up the phone. The remainder of this article will focus deeper on how tinyML works, and on current and future applications.

Image for post

The hierarchy of the cloud. (Source: eBizSolutions)

Examples of TinyML

Previously, complex circuitry was necessary for a device to perform a wide range of actions. Now, machine learning is making it increasingly possible to abstract such hardware “intelligence” into software, making embedded devices increasingly simple, lightweight, and flexible.

The challenges that machine learning with embedded devices presents are considerable, but great progress has already been achieved in this area. The key challenges in deploying neural networks on microcontrollers are the low memory footprint, limited power, and limited computation.

Perhaps the most obvious example of TinyML is within smartphones. These devices perpetually listen actively for ‘wake words’, such as “Hey Google” for Android smartphones, or ‘Hey Siri” on iPhones. Running these activities through the main central processing unit (CPU) of a smartphone, which is 1.85 GHz for the modern iPhone, would deplete the battery in just a few hours. This level of degradation is not acceptable for something that most people would use a few times a day at most.

To combat this, developers created specialized low-power hardware that is able to be powered by a small battery (such as a circular CR2032 “coin” battery). These allow the circuits to remain active even when the CPU is not running, which is basically whenever the screen is not lit.

These circuits can consume as little as 1 mW and can be powered for up to a year using a standard CR2032 battery.

It may not seem like it, but this is a big deal. Energy is a limiting factor for many electronic devices. Any device that requires mains electricity is restricted to locations with wiring, which can quickly get overwhelming when a dozen devices are present in the same location. Mains electricity is also inefficient and expensive. Converting mains voltage (which operates around 120 V in the United States) to a typical circuit voltage range (often ~5 V) wastes large amounts of energy. Anyone with a laptop charger will probably know this when unplugging their charger. The heat from the transformer within the charger is wasted energy during the voltage conversion process.

Even devices with batteries suffer from limited battery life, which requires frequent docking. Many consumer devices are designed such that the battery lasts for a single workday. TinyML devices that can continue operating for a year on a battery the size of a coin mean they can be placed in remote environments, only communicating when necessary in order to conserve energy.

Wake words are not the only TinyML we see seamlessly embedded in smartphones. Accelerometer data is used to determine whether someone has just picked the phone up, which wakes the CPU and turns on the screen.

Clearly, these are not the only possible applications of TinyML. In fact, TinyML presents many exciting opportunities for businesses and hobbyists alike to produce more intelligent IoT devices. In a world where data is becoming more and more important, the ability to distribute machine learning resources to memory-constrained devices in remote locations could have huge benefits on data-intensive industries such as farming, weather prediction, or seismology.

It is without a doubt that empowering edge devices with the capability of performing data-driven processing will produce a paradigm shift for industrial processes. As an example, devices that are able to monitor crops and send a “help” message when it detects characteristics such as soil moisture, specific gases (for example, apples emit ethane when ripe), or particular atmospheric conditions (e.g., high winds, low temperatures, or high humidity), would provide massive boosts to crop growth and hence crop yield.

As another example, a smart doorbell might be fitted with a camera that can use facial recognition to determine who is present. This could be used for security purposes, or even just so that the camera feed from the doorbell is fed to televisions in the house when someone is present so that the residents know who is at the door.

Two of the main focus areas of tinyML currently are:

Keyword spotting. Most people are already familiar with this application. “Hey Siri” and “Hey Google” are examples of keywords (often used synonymously with hotword or wake word). Such devices listen continuously to audio input from a microphone and are trained to only respond to specific sequences of sounds, which correspond with the learned keywords. These devices are simpler than automatic speech recognition (ASR) applications and utilize correspondingly fewer resources. Some devices, such as Google smartphones, utilize a cascade architecture to also provide speaker verification for security.

Visual Wake Words. There is an image-based analog to the wake words known as visual wake words. Think of these as a binary classification of an image to say that something is either present or not present. For example, a smart lighting system may be designed such that it activates when it detects the presence of a person and turns off when they leave. Similarly, wildlife photographers could use this to take pictures when a specific animal is present, or security cameras when they detect the presence of a person.

A more broad overview of current machine learning use cases of TinyML is shown below.

Image for post

Machine learning use cases of TinyML (Source Image: NXP).

How TinyML Works

TinyML algorithms work in much the same way as traditional machine learning models. Typically, the models are trained as usual on a user’s computer or in the cloud. Post-training is where the real tinyML work begins, in a process often referred to as deep compression.

Image for post

Diagram of the deep compression process. Source: ArXiv.

Model Distillation

Post-training, the model is then altered in such a way as to create a model with a more compact representation. Pruning and knowledge distillation are two such techniques for this purpose.

The idea underlying knowledge distillation is that larger networks have some sparsity or redundancy within them. While large networks have a high representational capacity, if the network capacity is not saturated it could be represented in a smaller network with a lower representation capacity (i.e., less neurons). Hinton et al. (2015) referred to the embedded information in the teacher model to be transferred to the student model as “dark knowledge”.

The below diagram illustrates the process of knowledge distillation.

Image for post

Diagram of the deep compression process. In this diagram, the ‘teacher’ is a trained convolutional neural network model. The teacher is tasked with transferring its ‘knowledge’ to a smaller convolutional network model with fewer parameters, the ‘student’. This process is known as knowledge distillation and is used to enshrine the same knowledge in a smaller network, providing a way of compressing networks such that they can be used on more memory-constrained devices. Source: ArXiv.

In this diagram, the ‘teacher’ is a trained neural network model. The teacher is tasked with transferring its ‘knowledge’ to a smaller network model with fewer parameters, the ‘student’. This process is used to enshrine the same knowledge in a smaller network, providing a way of compressing the knowledge representation, and hence the size, of a neural network such that they can be used on more memory-constrained devices.

Similarly, pruning can help to make the model’s representation more compact. Pruning, broadly speaking, attempts to remove neurons that provide little utility to the output prediction. This is often associated with small neural weights, whereas larger weights are kept due to their greater importance during inference. The network is then retrained on the pruned architecture to fine-tune the output.

Image for post

Illustration of pruning for distilling a model’s knowledge representation.

Quantization

Following distillation, the model is then quantized post-training into a format that is compatible with the architecture of the embedded device.

Why is quantization necessary? Imagine an Arduino Uno using an ATmega328P microcontroller, which uses 8-bit arithmetic. To run a model on the Uno, the model weights would ideally have to be stored as 8-bit integer values (whereas many desktop computers and laptops use 32-bit or 64-bit floating-point representation). By quantizing the model, the storage size of weights is reduced by a factor of 4 (for a quantization from 32-bit to 8-bit values), and the accuracy is often negligibly impacted (often around 1–3%).

Image for post

Illustration of quantization error during 8-bit encoding (which is then used to reconstruct 32-bit floats). (Source: TinyML book)

Some information may be lost during quantization due to quantization error (for example, a value that is 3.42 on a floating-point representation may be truncated to 3 on an integer-based platform). To combat this, quantization-aware (QA) training has also been proposed as an alternative. QA training essentially constrains the network during training to only use the values that will be available on the quantized device (see Tensorflow example).

Huffman Encoding

Encoding is an optional step that is sometimes taken to further reduce the model size by storing the data in a maximally efficient way: often via the famed Huffman encoding.

Compilation

Once the model has been quantized and encoded, it is converted to a format that can be interpreted by some form of light neural network interpreter, the most popular of which are probably TF Lite (~500 KB in size) and TF Lite Micro (~20 KB in size). The model is then compiled into C or C++ code (the languages most microcontrollers work in for efficient memory usage) and run by the interpreter on-device.

Image for post

The workflow of TInyML application (Source: TinyML book by Pete Warden and Daniel Situnayake)

Most of the skill of tinyML comes in dealing with the complex world of microcontrollers. TF Lite and TF Lite Micro are so small because any unnecessary functionality has been removed. Unfortunately, this includes useful abilities such as debugging and visualization. This means that it can be difficult to discern what is going on if there is an error during deployment.

Additionally, while the model has to be stored on the device, the model also has to be able to perform inference. This means the microcontroller must have a memory large enough that it can run (1) its operating system and libraries, (2) a neural network interpreter such as TF Lite, (3) the stored neural weights and neural architecture, and (4) the intermediate results during inference. Thus, the peak memory usage of a quantized algorithm is often quoted in tinyML research papers, along with memory usage, the number of multiply-accumulate units (MACs), accuracy, etc.

Why not train on-device?

Training on-device brings about additional complications. Due to reduced numerical precision, it becomes exceedingly difficult to guarantee the necessary level of accuracy to sufficiently train a network. Automatic differentiation methods on a standard desktop computer are approximately accurate to machine precision. Computing derivatives to the accuracy of 10^-16 is incredible, but utilizing automatic differentiation on 8-bit values will result in poor results. During backpropagation, these derivatives are compounded and eventually used to update neural parameters. With such a low numerical precision, the accuracy of such a model may be poor.

That being said, neural networks have been trained using 16-bit and 8-bit floating-point numbers.

The first paper looking at reducing numerical precision in deep learning was the 2015 paper Deep Learning with Limited Numerical Precision by Suyog Gupta and colleagues. The results of this paper were interesting, showing that the 32-bit floating-point representation could be reduced to a 16-bit fixed-point representation with essentially no degradation in accuracy. However, this is the only case when stochastic rounding is used because, on average, it produces an unbiased result.

In 2018, Naigang Wang and colleagues trained a neural network using 8-bit floating point numbers in their paper Training Deep Neural Networks with 8-bit Floating Point Numbers”. Training a neural network using 8-bit numbers rather than inference is significantly more challenging to achieve because of a need to maintain fidelity of gradient computations during backpropagation (which is able to achieve machine precision when using automatic differentiation).

How about compute-efficiency?

Models can also be tailored to make them more compute-efficient. Model architectures widely deployed on mobile devices such as MobileNetV1 and MobileNetV2 are good examples. These are essentially convolutional neural networks that have recast the convolution operation to make it more compute-efficient. This more efficient form of convolution is known as depthwise separable convolution. Architectures can also be optimized for latency using hardware-based profiling and neural architecture search, which are not covered in this article.

The Next AI Revolution

The ability to run machine learning models on resource-constrained devices opens up doors to many new possibilities. Developments may help to make standard machine learning more energy-efficient, which will help to quell concerns about the impact of data science on the environment. In addition, tinyML allows embedded devices to be endowed with new intelligence based on data-driven algorithms, which could be used for anything from preventative maintenance to detecting bird sounds in forests.

While some machine learning practitioners will undoubtedly continue to grow the size of models, a new trend is growing towards more memory-, compute-, and energy-efficient machine learning algorithms. TinyML is still in its nascent stages, and there are very few experts on the topic. I recommend the interested reader to examine some of the papers in the references, which are some of the important papers in the field of tinyML. This space is growing quickly and will become a new and important application of artificial intelligence in industry within the coming years. Watch this space.

Wolfe Systems is a Business Technology Solution provider. Our offerings are tailored to our clients requirements, we believe Technology is a tool that should be harnessed to assist businesses to achieve their goals. Find out how we can help you today. Contact us now!

Follow us on the WOLFE SYSTEMS socials for up to date tech trends,  information and cool facts

Top 10 Cyber Security Tips For Your Home Office and Business

Top 10 Cyber Security Tips For Your Home Office and Business

It’s frustrating that companies will pay a ransom and not take the time to hire a company ahead of time.

It’s much easier and cheaper to be preventative and to harden your system and be ready for attacks. 

Security is complex. That’s a fact. Here are a few tips YOU can do today!

Top 10 Cyber Security Tips

  • Reboot your mobile device(s) every morning
  • Use a microphone/camera blocker on all devices/computers when not in use
  • Use Multi-Factor Authentication (MFA) with all email/cloud/web accounts
  • Use a Password Manager (with strong passwords, no password reuse)
  • Use a Virtual Private Network (VPN), make sure the VPN vendor is based in a friendly country!
  • Make sure all devices/computers are fully patched (operating system/software/apps are always updated)
  • Don’t post addresses, phone numbers or email account information on social media
  • When travelling, don’t use airport/plane/hotel Wi-Fi networks unless absolutely necessary (and use a VPN if you do!)
  • At home, don’t use the Wi-Fi network provided by your ISP modem (use a separate Wi-Fi router)
  • Keep home IoT (smart speakers, TVs, etc) on a separate Wi-Fi network from devices/computers
WOLFE SYSTEMS Can Assess Your Current Risk

We will run a FREE  Scan for your Organisation, to evaluate your current Risk Posture.

On going Monitoring will allow you to know when there are new exposures associated with your domain, so you can take the actions to close the doors before a criminal accesses your network, your data, your business.

If you don’t monitor, you won’t know…

until it’s too late!




Contact us today on 1300 958 923 to discover our findings and set up a 30min introductory meeting


Wolfe Systems is a Business Technology Solution provider. Our offerings are tailored to our clients requirements, we believe Technology is a tool that should be harnessed to assist businesses to achieve their goals. Find out how we can help you today. Contact us now!

Follow us on the WOLFE SYSTEMS socials for up to date tech trends,  information and cool facts

Rethinking cyber security for remote working

Rethinking cyber security for remote working

Fireside Chat Screenshot
CSO and Mimecast Webinar
The COVID-19 pandemic has precipitated a huge increase in remote working, posing numerous challenges for management and for individuals. Arguably the greatest is the cyber security challenge created by organisations needing to speedily grant access to corporate systems from many devices beyond the direct control of CISOs and their security teams.It’s expected that the current level of remote working will reduce over time, but not to pre-pandemic levels. Future-of-work research firm Global Workplace Analytics estimates 25-30 percent of the workforce will work from home on multiple days each week by the end of 2021.

The shift to home working has created two distinct cyber security challenges: maintaining security through the rapid transition, and securing IT in a future where remote working is the norm rather than the exception.

Meanwhile the bad actors have lost no time in exploiting the opportunities presented by remote working. Mimecast detected a 30+ percent increase in each of spam, impersonation, malware and suspicious domains in the three months following 31 March 2020.

Rethink your security posture

To identify the cyber security challenges of sustained remote working, and offer solutions, Mimecast sponsored a webinar, Why your cyber security posture needs a rethink.

CSO associate editor, Byron Connolly, chaired a panel comprising cyber security analyst James Turner from CISO Lens, Garrett O’Hara, Principle Technical Consultant with Mimecast, and Chris Neal, CISO at Ramsay Health Care.

Turner set the scene for the panel discussion by identifying three forces driving the need to rethink organisations’ cyber security: increased risk, economics and geopolitics.

Summing up the risks created by the surge in demand for home working, he said: “A lot of risks were just accepted in the rush to do it. And now, CISOs and CIOs are going back over the risks they’ve accepted over the past several months and asking if it is still appropriate to be accepting those risks.

“How do we best enshrine those processes given all the indicators are that this is going to go on for at least the next couple of years?”

It was evident compromises had been made to balance cyber security and operational priorities. Neal said, in Ramsay Health Care, that threat awareness training had been wound back.

“If it’s a choice between a nurse caring for patients or trying to deal with COVID patients versus spending five minutes on a security awareness video, I know which that has to be.”

More seriously, O’Hara predicted that many cyber security decisions taken under pressure from the pandemic would create problems down the track. “I see a big piece of work in 12 to 18 months where people go ‘Oh my god! All this stuff has happened. How do we get the toothpaste back in the tube?’”

Meeting the cyber skills shortage

To add to the challenges, the cyber security demands engendered by home working have exacerbated an already serious shortage of cyber security skills, and the discussion turned to how this problem might be addressed.

Turner said there was a growing trend to fill security roles from other areas of IT. “The CISO community is looking to train existing technologists to care more about security themselves. If we can shift their understanding so they get how what they do has a direct impact on security … we’ve got better security practices and thinking coming from people that already understand the tech.”

Neal confirmed this approach saying he had hired no cyber security specialists. “Most of my team I’ve poached from other parts of Ramsay Health Care IT: people who knew how Ramsey worked, knew the technology, knew the business and had an interest in and an aptitude for cyber.”

Three part solution to security challenges

Each of the three experts in the webinar — analyst, vendor, user — brought a different perspective to the COVID-19 induced cyber security challenge, but all agreed that meeting this challenge required an effective combination of cyber security skills, technology skills, and business and communication skills.

O’Hara said having good, well-integrated cyber security platforms would free up cyber security personnel for more meaningful roles.

Turner said security staff must understand the businesses they were hired to protect. “Security people need to have an intimate understanding of how the business uses technology, but they can’t know everything. They are completely dependent on their communication abilities with both the business and with IT.”

Neal said drawing cyber security staff from other areas of IT had a dual benefit: they knew the business and could educate the business about the importance of cyber security.

You’ll find these and many more valuable insights from the front line of cyber security in a COVID-19 world in our webinar. Why your cyber security posture needs a rethink. Watch it here.

You can find the original article here

WOLFE SYSTEMS Can Assess Your Current Risk

We will run a FREE  Scan for your Organisation, to evaluate your current Risk Posture.

On going Monitoring will allow you to know when there are new exposures associated with your domain, so you can take the actions to close the doors before a criminal accesses your network, your data, your business.

If you don’t monitor, you won’t know…

until it’s too late!




Contact us today on 1300 958 923 to discover our findings and set up a 30min introductory meeting


Wolfe Systems is a Business Technology Solution provider. Our offerings are tailored to our clients requirements, we believe Technology is a tool that should be harnessed to assist businesses to achieve their goals. Find out how we can help you today. Contact us now!

Follow us on the WOLFE SYSTEMS socials for up to date tech trends,  information and cool facts