Articles Featured Videos
Artificial Intelligence: Transferring biases to machines
February 2, 2018
, , , ,

Are we transferring our biases to computers? With supervised learning, machines learn what we teach them. Just like our kids often perpetuate our belief systems, biases, and preferences, will machine also carry forward our implicit biases?

Leading research on prejudice, the study of implicit social cognition, highlight the distinction between “controlled” and “automatic” information processing. Cognitive psychologists Shiffrin & Schneider in their seminal paper “Controlled and automatic human information processing: Perceptual learning, automatic attending, and a general theory,” significantly advanced the implicit bias theory. There has been a consensus that:

  • Controlled processing is “voluntary, attention-demanding, and of limited capacity.”
  • Automatic processing occurs “without attention,” has “nearly unlimited capacity,” and “hard to suppress voluntarily.”

Growing body of knowledge on implicit bias and credible studies show that awareness of stereotypes can affect social judgment and behavior in relative independence from how subjects respond to measures of their explicit attitudes. Simply put, an egalitarian, well-behaved man can display semi-sexist tendencies.

Machine learning is fundamental to artificial intelligence, or A.I. In supervised learning, computers learn like a human. AI “modeled on science’s understanding of how the human brain processes and categorizes information.”

In supervised learning, if we teach machines that name “Ibrahim” or “Mohammad” pose a higher security risk, will machine become Islamophobic?

Watch this Facebook Brand Studio production on Artificial Intelligence.


According to the U.S. Bureau of Justice Statistics (BJS) in 2013 black males accounted for 37% of the total male prison population, white males 32%, and Hispanic males 22%. White females comprised 49% of the prison population in comparison to black females who accounted for 22% of the female population. However, the imprisonment rate for black females (113 per 100,000) was 2x the rate for white females (51 per 100,000). Out of all ethnic groups, African Americans, Puerto Rican Americans, and American Indians have some of the highest rates of incarceration. Though, of these groups, the black population is the largest, and therefore make up a large portion of those incarcerated in US prisons and jails. When I look at the following chart, I wonder, if machines are learning that “Black” and “Hispanic” pose a higher risk to the society?

2010. Inmates in adult facilities, by race and ethnicity. Jails, and state and federal prisons.
Race, ethnicity % of US population % of U.S.
incarcerated population
National incarceration rate
(per 100,000 of all ages)
White (non-Hispanic) 64 39 450 per 100,000
Hispanic 16 19 831 per 100,000
Black 13 40 2,306 per 100,000
Articles Featured
TensorFlow is gaining currency!
January 31, 2018

This rock-paper-scissors science experiment is a novel use of TensorFlow.  Last year TensorFlow was seen automating cucumber sorting, finding sea cows in aerial imagery, sorting diced potatoes to make safer baby food, identifying skin cancer, helping to interpret bird call recordings in a New Zealand bird sanctuary, and identifying diseased plants in the most popular root crop on Earth in Tanzania!  #AI #ArtificialIntelligence

Training state-of-the-art machine learning models requires an enormous amount of computation, and researchers, engineers, and data scientists often wait weeks for results. To solve this problem, Google “designed an all-new ML accelerator from scratch — a second-generation TPU, or Tensor Processing Unit — that can accelerate both training and running ML models.”

Each device delivers up to 180 teraflops of floating-point performance, and these new TPUs are designed to be connected into even larger systems. A 64-TPU pod can apply up to 11.5 petaflops of computation to a single ML training task.

We’re extremely excited about these new TPUs, and we want to share this technology with the world so that everyone can access their benefits. That’s why we’re bringing our second-generation TPUs to Google Cloud for the first time as Cloud TPUs on GCE, the Google Compute Engine. You’ll be able to mix-and-match Cloud TPUs with Skylake CPUs, NVIDIA GPUs, and all of the rest of our infrastructure and services to build and optimize the perfect machine learning system for your needs. Best of all, Cloud TPUs are easy to program via TensorFlow, the most popular open-source machine learning framework.

Google’s new Cloud TPUs deliver up to 180 teraflops of machine learning acceleration
Amazon-Google Feud
December 6, 2017

Are you following Google-Amazon feud?

Google has decided to remove YouTube from all Fire TV products and the Echo Show apparently after Amazon’s decided to delist new Nest products.

Amazon claims that “Google is setting a disappointing precedent by selectively blocking customer access to an open website.”

Amazon has also refused to sell Google products such as Chromecast and Google Home. Fire TV owners will lose YouTube January 1.

Should Google pull YouTube from Amazon hardware? #GoogleAmazonFeud

Articles Featured
Privacy: Can WiFi data be anonymous?
October 19, 2017
, ,

While in the past, mobility traces were only available to mobile phone carriers, the advent of smartphones and other means of data collection has made these broadly available.

Last year Transport for London (TfL) launched trial to gather “de-personalised WiFi connection data collected at 54 London Underground stations within Zones 1-4 to help improve the services it offers customers.”

But TechCrunch reports TfL has now turned down an FOI request asking for it to release the “full dataset of anonymized data for the London Underground Wifi Tracking Trial” — arguing that it can’t release the data as there is a risk of individuals being re-identified (and disclosing personal data would be a breach of UK data protection law).

“Although the MAC address data has been pseudonymised, personal data as defined under the [UK] Data Protection Act 1998 is data which relate to a living individual who can be identified from the data, or from those data and other information which is in the possession of, or is likely to come into the possession of, the data controller,” TfL writes in the FOI response in which it refuses to release the dataset.

A simply anonymized dataset does not contain name, home address, phone number or other obvious identifier. Yet, if individual’s patterns are unique enough, outside information can be used to link the data back to an individual. See below :

(A) Trace of an anonymized mobile phone user during a day.The dots represent the times and locations where the user made or received a call. Every time the user has such an interaction, the closest antenna that routes the call is recorded.

(B) The same user’s trace as recorded in a mobility database. The Voronoi lattice, represented by the grey lines, is an approximation of the antennas reception areas, the most precise location information available to us. The user’s interaction times are here recorded with a precision of one hour.

(C) The same individual’s trace when we lower the resolution of our dataset through spatial and temporal aggregation. Antennas are aggregated in clusters of size two, and their associated regions are merged. The user’s interaction is recorded with a precision of two hours. Such spatial and temporal aggregation render the 8:32 am and 9:15 am interactions indistinguishable.


Articles Featured
KRACK Attack: Vulnerabilities in Wi-Fi Protected Access and Wi-Fi Protected Access II
October 17, 2017
, , , ,

We are all exposed to the Krack Wi-Fi security vulnerability—a flaw that puts any person using wireless internet at risk of being hacked.

The Krack security exploit was discovered by Mathy Vanhoef, a cybersecurity expert at Belgian university KU Leuven, who will present his research at the Computer and Communications Security (CCS) conference later this month.

“We discovered serious weaknesses in WPA2, a protocol that secures all modern protected WiFi networks,” Vanhoef wrote in a blogpost describing the vulnerability. “An attacker within range of a victim can exploit these weaknesses using key reinstallation attacks (KRACKs). Concretely, attackers can use this novel attack technique to read information that was previously assumed to be safely encrypted.”

“This implies all these networks are affected by (some variant of) our attack. For instance, the attack works against personal and enterprise Wi-Fi networks, against the older WPA and the latest WPA2 standard, and even against networks that only use AES.”


Multiple Cisco wireless products are affected by these vulnerabilities

In a statement, today Cisco acknowledged multiple wireless products are affected by these vulnerabilities and said it will release software updates to address these vulnerabilities. There is a workaround that addresses the vulnerability in CVE-2017-13082. There are no workarounds that address the other vulnerabilities described in this advisory.

This advisory is available at the following link:

Apple also claims to have fixed the issue in certain versions of its operating systems, including iOS used on iPhones and watch OS used on the Apple Watch, and macOS used on Apple Macs. The patches, however, are mostly available only for trial versions of the software and therefore are available only for developers.

 “Microsoft released security updates on October 19 and customers who have Windows Update enabled and applied the security updates are protected automatically,” the company said in a statement. “We updated to protect customers as soon as possible, but as a responsible industry partner, we withheld disclosure until other vendors could develop and release updates.”


Google has yet to issue any fixes for the Krack attack method, saying in a statement on Monday that it is working on ways to resolve it.

A research paper with the title of “Key Reinstallation Attacks: Forcing Nonce Reuse in WPA2” was made publicly available. This paper discusses seven vulnerabilities affecting session key negotiation in both the Wi-Fi Protected Access (WPA) and the Wi-Fi Protected Access II (WPA2) protocols. These vulnerabilities may allow the reinstallation of a pairwise transient key, a group key, or an integrity key on either a wireless client or a wireless access point. Additional research also led to the discovery of three additional vulnerabilities (not discussed in the original paper) affecting wireless supplicant supporting either the 802.11z (Extensions to Direct-Link Setup) standard or the 802.11v (Wireless Network Management) standard. The three additional vulnerabilities could also allow the reinstallation of a pairwise key, group key, or integrity group key.

‘My God, it’s better’: Emma can write again thanks to a prototype watch, raising hope for Parkinson’s disease – Transform
October 6, 2017

Microsoft researcher Haiyan Zhang created a watch for Emma Lawton that helps the graphic designer control the symptoms of Parkinson’s disease.

Sourced through from: