Representation & Reasoning


The Simple Math behind 3 Decision Tree Splitting criterions

#artificialintelligence

Decision Trees are great and are useful for a variety of tasks. They form the backbone of most of the best performing models in the industry like XGboost and Lightgbm. But how do they work exactly? In fact, this is one of the most asked questions in ML/DS interviews. We generally know they work in a stepwise manner and have a tree structure where we split a node using some feature on some criterion.


Software robots' workforce contributions will increase 50% in the next 2 years

#artificialintelligence

No, the robots are not coming for your job as they ready to take over the world ... yet. But the future of the world's workforce will mark a significant shift and work will be heavily reliant on the teamwork of human and machine, noted the just-released IDC white paper, Content Intelligence for the Future of Work. And we're not quite in sci-fi film territory either, said Holly Muscolino, research vice president of content and process strategies and the future of work at IDC. "A software robot (or'digital worker') is essentially a software program that automates a task that has previously been accomplished by a human worker," Muscolino explained. "The term'robot' is used to signify the role that these software solutions play in automation, however, beyond that, there is no relationship between a software robot and the physical robots that we may see on the manufacturing line, patrolling supermarket aisles on starring in'Star Wars'' movies." Muscolino added, "A variety of software technologies are classified as'digital workers.' The technology gaining the most airtime today is robotic process automation (RPA), but other automation technologies, and AI-enabled technologies, like digital assistants and chatbots, are also classified as'digital workers'."


Rethinking Search Engines and Recommendation Systems

Communications of the ACM

In her popular book, Weapons of Math Destruction, data scientist Cathy O'Neil elegantly describes to the general population the danger of the data science revolution in decision making. She describes how the US News ranking of universities, which orders universities based on 15 measured properties, created new dynamics in university behavior, as they adapted to these measures, ultimately resulting in decreased social welfare. Unfortunately, the idea that data science-related algorithms, such as ranking, cause changes in behavior, and that this dynamic may lead to socially inferior outcomes, is dominant in our new online economy. Ranking also plays a crucial role in search engines and recommendation systems--two prominent data science applications that we focus on in this article. Recommendation systems endorse items by ranking them using information induced from some context--for example, the Web page a user is currently browsing, a specific application the user is running on her mobile phone, or the time of day.


Automated Program Repair

Communications of the ACM

For the running example in Figure 1, this abstraction would replace the application-specific identifiers triangle and EQUILATERAL with generic placeholders, such as VAR1 and VAR2. After this abstraction, both approaches use an RNN-based sequence-to-sequence network that predicts how to modify the abstracted code. Given the increasing interest in learning-based approaches toward software engineering problems, we will likely see more progress on learning-based repair in the coming years. Key challenges toward effective solutions include finding an appropriate representation of source code changes and obtaining large amounts of high-quality human patches as training data.


Uncertainty

Communications of the ACM

In a famous episode in the "I Love Lucy" television series--"Job Switching," better known as the chocolate factory episode--Lucy and her best-friend coworker Ethel are tasked to wrap chocolates flowing by on a conveyor belt in front of them. Each time they get better at the task, the conveyor belt speeds up. Eventually they cannot keep up and the whole scene collapses into chaos. The threshold between order and chaos seems thin. A small perturbation--such as a slight increase in the speed of Lucy's conveyor belt--can either do nothing or it can trigger an avalanche of disorder. The speed of events within an avalanche overwhelms us, sweeps away structures that preserve order, and robs our ability to function.


Google Seeks People With Down Syndrome To Help Train AIs To Understand Human Speech

#artificialintelligence

The last half decade has ushered in the era of humans interacting with technology through speech, with Amazon's Alexa, Apple's Siri, and Google's AI rapidly becoming ubiquitous elements of the human experience. But, while the migration from typing to voice has brought great convenience for some folks (and improved safety, in the case of people utilizing technology while driving), it has not delivered on its potential for the people who might otherwise stand to benefit the most from it: those of us with disabilities. For people with Down Syndrome, for example, voice-based control of technology offers the promise of increased independence – and even of some new, potentially life-saving products. Yet, for this particular group of people, today's voice-recognizing AIs pose serious problems, as a result of a combination of 3 factors: To address this issue, and as a step forward towards ensuring that people with health conditions that cause AIs to be unable to understand them are able to utilize modern technology, Google is partnering with the Canadian Down Syndrome Society; via an effort called Project Understood, Google hopes to obtain recordings of people with Down Syndrome reading simple phrases, and to use those recordings to help train its AI to understand the speech patterns common to those with Down Syndrome. This effort is an extension of Google's own Project Euphonia, which seeks to improve computers' abilities to understand diverse speech patterns including impaired speech, and, which, earlier this year, began an effort to train AIs to recognize communication from people with the neuro-degenerative condition ALS, commonly known as Lou Gehrig's Disease.


The Human-Centric Rise of Artificial Intelligence in Healthcare Lenovo StoryHub

#artificialintelligence

Dressing one morning, as she usually does, Jane notices a strange skin discoloration on her arm. Still smaller than a dime, but she swears it used to be half that size and certainly more symmetrical. She asks her virtual assistant to scan the area and assess. A camera built into her bathroom mirror fires up, captures photos, and checks them against archival images from Jane's entire photo library. Jane was right; something is wrong.


The best Cyber Monday deals according to Alexa: any Amazon-owned brand

#artificialintelligence

Cyber Monday was the biggest shopping day in Amazon's history. Turns out it was also a day for Amazon to promote its own brands over all others. On Monday, asking an Amazon Echo for the "best Cyber Monday deals" returned five straight responses from smart assistant Alexa that promoted products owned by Amazon: Amazon acquired Blink, a maker of security cameras, in 2017; it purchased Ring, the maker of video doorbells, in 2018. An Amazon spokesperson told Recode that after those first five deals, Amazon did promote items made by other companies, and that Alexa-enabled non-Echo devices returned deals in their first five responses that included non-Amazon-owned items. But the initial Alexa responses are just another sign of Amazon's aggressiveness in promoting its own brands, whether they are gadgets like the Echo family of smart speaker products and Ring doorbells or apparel lines like Goodthreads.


MIT creates an AI that understands the laws of physics intuitively - Innowire

#artificialintelligence

Dubbed ADEPT, the system is able to, like a human being, understand some laws of physics intuitively. It can look at an object in a video, predict how it should act based on what it knows of the laws of physics and then register surprise if what it was looking at subsequently vanishes or teleports. The team behind ADEPT say their model will allow other researchers to create smarter AIs in the future, as well give us a better understanding of how infants understand the world around them. "By the time infants are three months old, they have some notion that objects don't wink in and out of existence, and can't move through each other or teleport," said Kevin A. Smith, one of the researchers that created ADEPT. "We wanted to capture and formalize that knowledge to build infant cognition into artificial-intelligence agents. We're now getting near human-like in the way models can pick apart basic implausible or plausible scenes."


Would background checks make dating apps safer?

#artificialintelligence

Match Group, the largest dating app conglomerate in the US, doesn't perform background checks on any of its apps' free users. A ProPublica report today highlights a few incidents in which registered sex offenders went on dates with women who had no idea they were talking to a convicted criminal. These men then raped the women on their dates, leaving the women to report them to the police and to the apps' moderators. These women expected their dating apps to protect them, or at least vet users, only to discover that Match has little to no insight on who's using their apps. The piece walks through individual attacks and argues that the apps have no real case for not vetting their users.