Why we need XAI, not just responsible AI
There are many techniques organisations can use to develop XAI. As well as
continually teaching their system new things, they need to ensure that it is
learning correct information and does not use one mistake or piece of biased
information as the basis for all future analysis. Multilingual semantic
searches are vital, particularly for unstructured information. They can filter
out the white noise and minimise the risk of seeing the same risk or
opportunity multiple times. Organisations should also add a human element to
their AI, particularly if building a watch list. If a system automatically red
flags criminal convictions without scoring them for severity, a person with a
speeding fine could be treated in the same way as one serving a long prison
sentence. For XAI, systems should always err on the side of the positive. If a
red flag is raised, the AI system should not give a flat ‘no’ but should raise
an alert for checking by a human. Finally, even the best AI system should
generate a few mistakes. Performance should be an eight out of ten, never a
ten, or it becomes impossible to trust that the system is working properly.
Mistakes can be addressed, and performance continually tweaked, but there will
never be perfection.
What classic software developers need to know about quantum computing
There are many different parts of quantum that are exciting to study. One is
quantum computing using quantum to do any sort of information processing, the
other is communication itself. And maybe the third part that doesn't get as
much media attention but should is sensing, using quantum computers to sense
things much more sensitively than you would classically. So think about
sensing very small magnetic fields for example. So the communication aspect of
it is just as important because at the end of the day it's important to have
secure communication between your quantum computers as well. So this is
something exciting to look forward to. ... So the first tool that you need,
and one of the most important tools is the one that gives you access to the
quantum computers. So if you go to quantum-computing.ibm.com and create an
account there, we give you immediate access to several quantum computers,
which first of all, every time I say, this just blows my mind because four
years ago this wasn't a thing. You couldn't go online and access a quantum
computer. I was in grad school because I wanted to do quantum research and
needed access to a lab to do this work
Why Darknet Markets Persist
"There are two main reasons here: the lack of alternatives and the ease of use
of marketplaces," researchers at the Photon Research Team at digital risk
protection firm Digital Shadows tell Information Security Media Group. At
least for English-speaking users, such considerations often appear to trump
other options, which include encrypted messaging apps as well as forums
devoted to cybercrime or hacking. And many users continue to rely on markets
despite the threat of exit scams, getting scammed by sellers or getting
identified and arrested by police. Another option is Russian-language
cybercrime forums, which continue to thrive, with many hosting high-value
items. But researchers say that, even when armed with translation software,
English speakers often have difficulty coping with Russian cybercrime argot.
Many Russian speakers also refuse to do business with anyone from the West.
... Demand for new English-language cybercrime markets continues to be high
because so many existing markets get disrupted by law enforcement agencies or
have administrators who run an exit scam. Before Empire, other markets that
closed after their admins "exit scammed" have included BitBazaar in August,
Apollon in March and Nightmare in August 2019.
Open Data Institute explores diverse range of data governance structures
The involvement of different kinds of stakeholders in any particular
institution also has an effect on what kinds of governance structures would be
appropriate, as different incentives are needed to motivate different actors
to behave as responsible and ethical stewards of the data. In the context of
the private sector, for example, enterprises that would normally adopt a
cut-throat, competitive mindset need to be incentivised for collaboration.
Meanwhile, cash-strapped third-sector organisations, such as charities and
non-governmental organisations (NGOs), need more financial backing to realise
the potential benefits of data institutions. “Many [private sector]
organisations are well-versed in stewarding data for their own benefit, so
part of the challenge here is for existing data institutions in the private
sector to steward it in ways that unlock value for other actors, whether
that’s economic value from say a competition point of view, but then also from
a societal point of view,” said Hardinges. “Getting organisations to consider
themselves data institutions, and in ways that unlock public value from
private data, is a really important part of it.”
5 supply chain cybersecurity risks and best practices
Falling prey to the "it couldn't happen to us" mentality is a big mistake. But
despite clear evidence that supply chain cyber attacks are on the rise, some
leaders aren't facing that reality, even if they do understand techniques to
build supply chain resilience more broadly. One of the biggest supply chain
challenges is leaders thinking they're not going to be hacked, said Jorge Rey,
the principal in charge of information security and compliance for services at
Kaufman Rossin, a CPA and advisory firm in Miami. To fully address supply
chain cybersecurity, supply chain leaders must realize they need to face the
risk reality. The supply chain is veritable smorgasbord of exploit
opportunities -- there are so many information and product handoffs in even a
simpler one -- and each handoff represents risks, especially where digital
technology is involved but easily overlooked. ... Supply chain cyber attacks
are carried out with different goals in mind -- from ransom to sabotage to
theft of intellectual property, Atwood said. These cyberattacks can also take
many forms, such as hijacking software updates and injecting malicious code
into legitimate software, as well as targeting IT and operational technology
and hitting every domain and any node, Atwood said.
Moving Toward Smarter Data: Graph Databases and Machine Learning
Data plays a significant role in machine learning, and formatting it in ways
that a machine learning algorithm can train on is imperative. Data pipelines
were created to address this. A data pipeline is a process through which raw
data is extracted from the database (or other data sources), is transformed,
and is then loaded into a form that a machine learning algorithm can train and
test on. Connected features are those features that are inherent in the
topology of the graph. For example, how many edges (i.e., relationships) to
other nodes does a specific node have? If many nodes are close together in the
graph, a community of nodes may exist there. Some nodes will be part of that
community while others may not. If a specific node has many outgoing
relationships, that node’s influence on other nodes could be higher, given the
right domain and context. Like other features being extracted from the data
and used for training and testing, connected features can be extracted by
doing a custom query based on the understanding of the problem space. However,
given that these patterns can be generalized for all graphs, unsupervised
algorithms have been created that extract key information about the topology
of your graph data and used as features for training your model.
Dark Side of AI: How to Make Artificial Intelligence Trustworthy
Malicious inputs to AI models can come in the form of adversarial AI,
manipulated digital inputs or malicious physical inputs. Adversarial AI may
come in the form of socially engineering humans using an AI-generated voice,
which can be used for any type of crime and considered a “new” form of
phishing. For example, in March of last year, criminals used AI synthetic
voice to impersonate a CEO’s voice and demand a fraudulent transfer of
$243,000 to their own accounts. Query attacks involve criminals sending
queries to organizations’ AI models to figure out how it's working and may
come in the form of a black box or white box. Specifically, a black box query
attack determines the uncommon, perturbated inputs to use for a desired
output, such as financial gain or avoiding detection. Some academics have been
able to fool leading translation models by manipulating the output, resulting
in an incorrect translation. A white box query attack regenerates a training
dataset to reproduce a similar model, which might result in valuable data
being stolen. An example of such was when a voice recognition vendor fell
victim to a new, foreign vendor counterfeiting their technology and then
selling it, which resulted in the foreign vendor being able to capture market
share based on stolen IP.
DDoS attacks rise in intensity, sophistication and volume
The total number of attacks increased by over two and a half times during
January through June of 2020 compared to the same period in 2019. The increase
was felt across all size categories, with the biggest growth happening at
opposite ends of the scale – the number of attacks sized 100 Gbps and above
grew a whopping 275% and the number of very small attacks, sized 5 Gbps and
below, increased by more than 200%. Overall, small attacks sized 5 Gbps and
below represented 70% of all attacks mitigated between January and June of
2020. “While large volumetric attacks capture attention and headlines, bad
actors increasingly recognise the value of striking at low enough volume to
bypass the traffic thresholds that would trigger mitigation to degrade
performance or precision target vulnerable infrastructure like a VPN,” said
Michael Kaczmarek, Neustar VP of Security Products. “These shifts put every
organization with an internet presence at risk of a DDoS attack – a threat
that is particularly critical with global workforces reliant on VPNs for
remote login. VPN servers are often left vulnerable, making it simple for
cybercriminals to take an entire workforce offline with a targeted DDoS
attack.”
Group Privacy and Data Trusts: A New Frontier for Data Governance?
The concept of collective privacy shifts the focus from an individual
controlling their privacy rights, to a group or a community having data rights
as a whole. In the age of Big Data analytics, the NPD Report does well to
discuss the risks of collective privacy harms to groups of people or
communities. It is essential to look beyond traditional notions of privacy
centered around an individual, as Big Data analytical tools rarely focus on
individuals, but on drawing insights at the group level, or on “the crowd” of
technology users. In a revealing example from 2013, data processors who
accessed New York City’s taxi trip data (including trip dates and times) were
able to infer with a degree of accuracy whether a taxi driver was a devout
Muslim or not, even though data on the taxi licenses and medallion numbers had
been anonymised. Data processors linked pauses in taxi trips with adherence to
regularly timed prayer timings to arrive at their conclusion. Such findings
and classifications may result in heightened surveillance or discrimination
for such groups or communities as a whole. .... It might be in the interest of
such a community to keep details about their ailment and residence private, as
even anonymised data pointing to their general whereabouts could lead to
harassment and the violation of their privacy.
Analysis: Online Attacks Hit Education Sector Worldwide
The U.S. faces a rise in distributed denial-of-service attacks, while Europe
is seeing an increase in information disclosures attempts - many of them
resulting from ransomware incidents, the researchers say. Meanwhile, in Asia,
cybercriminals are taking advantage of vulnerabilities in the IT systems that
support schools and universities to wage a variety of attacks. DDoS and other
attacks are surging because threat actors see an opportunity to disrupt
schools resuming online education and potentially earn a ransom for ending an
attack, according to Check Point and other security researchers. "Distributed
denial-of-service attacks are on the rise and a major cause of network
downtime," the new Check Point report notes. "Whether executed by hacktivists
to draw attention to a cause, fraudsters trying to illegally obtain data or
funds or a result of geopolitical events, DDoS attacks are a destructive cyber
weapon. Beyond education and research, organizations from across all sectors
face such attacks daily." In the U.S., the Cybersecurity and Infrastructure
Security Agency has warned of an increase in targeted DDoS attacks against
financial organizations and government agencies
Quote for the day:
"One of the most sincere forms of respect is actually listening to what another has to say." -- Bryant H. McGill
No comments:
Post a Comment