Existing encryption systems rely on specific mathematical equations that classical computers aren’t very good at solving — but quantum computers may breeze through them. As a security researcher, Chen is particularly interested in quantum computing’s ability to solve two types of math problems: factoring large numbers and solving discrete logarithms. Pretty much all internet security relies on this math to encrypt information or authenticate users in protocols such as Transport Layer Security. These math problems are simple to perform in one direction, but difficult in reverse, and thus ideal for a cryptographic scheme. “From a classical computer’s point of view, these are hard problems,” says Chen. “However, they are not too hard for quantum computers.” In 1994, the mathematician Peter Shor outlined in a paper how a future quantum computer could solve both the factoring and discrete logarithm problems, but engineers are still struggling to make quantum systems work in practice. While several companies like Google and IBM, along with startups such as IonQ and Xanadu, have built small prototypes, these devices cannot perform consistently, and they have not conclusively completed any useful task beyond what the best conventional computers can achieve.
Up to now, serverless technology has not been able to support stateful, high-performance, scalable applications that enterprises are building today, Murdoch said. Examples of such applications include consumer and industrial IoT, factory automation, modern e-commerce, real-time financial services, streaming media, internet-based gaming and SaaS applications. “Stateful approaches to serverless application design will be required to support a wide range of enterprise applications that can’t currently take advantage of it, such as e-commerce, workflows and anything requiring a human action,” said William Fellows, research director for cloud native at 451 Research. “Serverless functions are short-lived and lose any ‘state’ or context information when they execute.” Lightbend, with Akka Serverless, has addressed the challenge of managing distributed state at scale. “The most significant piece of feedback that we’ve been getting from the beta is that one of the key things that we had to do to build this platform was to find a way to be able to make the data be available in memory at runtime automatically, without the developer having to do anything,” Murdoch said
While challenges like these often sound theoretical, they already affect and shape the work that machine learning engineers and researchers produce. Angela Shi looks at a practical application of this conundrum when she explains the visual representation of bias and variance in bulls-eye diagrams. Taking a few steps back, Federico Bianchi and Dirk Hovy’s article identifies the most pressing issues the authors and their colleagues face in the field of natural learning processing (NLP): “the speed with which models are published and then used in applications can exceed the discovery of their risks and limitations. And as their size grows, it becomes harder to reproduce these models to discover those aspects.” Federico and Dirk’s post stops short of offering concrete solutions—no single paper could—but it underscores the importance of learning, asking the right (and often most difficult) questions, and refusing to accept an untenable status quo. If what inspires you to take action is expanding your knowledge and growing your skill set, we have some great options for you to choose from this week, too.
While agility might be critical for sporting success, that doesn't mean it's easily achieved. Filippi tells ZDNet he's spent many years building a strong team, with great heads of department who are empowered to make big calls. "Most of the time you trust them to get on with it," he says. "I'm more of an orchestrator – you cannot micromanage a race team because there's just too much going on. The pace and the volume of work being achieved every week is just mind-blowing." Hackland has similar experiences at Williams F1. Employees are empowered to take decisions and their confidence to make those calls in the factory or out on the track is a crucial component of success. "The engineer who's sitting on the pit wall doesn't have to ask the CIO if we should pit," he says. "The decisions that are made all through the organisation don't feed up to one single individual. Everyone is allowed to make decisions up or down the organisation." As well as being empowered to make big calls, Hackland says a no-blame culture is critical to establishing and supporting decentralised decision making in racing teams.
Disconnects also exist between key functional stakeholders required to make sound holistic judgements around ethics in AI and ML. “There is a gap between the bit that is the data analytics AI, and the bit that is the making of the decision by an organisation. You can have really good technology and AI generating really good outputs that are then used really badly by humans, and as a result, this leads to really poor outcomes,” says Prof. Leonard. “So, you have to look not only at what the technology in the AI is doing, but how that is integrated into the making of the decision by an organisation.” This problem exists in many fields. One field in which it is particularly prevalent is digital advertising. Chief marketing officers, for example, determine marketing strategies that are dependent upon the use of advertising technology – which are in turn managed by a technology team. Separate to this is data privacy which is managed by a different team, and Prof. Leonard says each of these teams don’t speak the same language as each other in order to arrive at a strategically cohesive decision.
As data scientists, the first and foremost skill we need is to think in terms of models. In its most abstract form, a model is any physical, mathematical, or logical representation of an object, property, or process. Let’s say we want to build an aircraft engine that will lift heavy loads. Before we build the complete aircraft engine, we might build a miniature model to test the engine for a variety of properties (e.g., fuel consumption, power) under different conditions (e.g., headwind, impact with objects). Even before we build a miniature model, we might build a 3-D digital model that can predict what will happen to the miniature model built out of different materials. ... Data scientists often approach problems with cross-sectional data at a point in time to make predictions or inferences. Unfortunately, given the constantly changing context around most problems, very few things can be analyzed statically. Static thinking reinforces the ‘one-and-done’ approach to model building that is misleading at best and disastrous at its worst. Even simple recommendation engines and chatbots trained on historical data need to be updated on a regular basis.
Over the past 12 months, double extortion attacks have become increasingly common as its ‘business model’ has proven effective. The data center giant Equinix was hit by the Netwalker ransomware. The threat actor behind that attack was also responsible for the attack against K-Electric, the largest power supplier in Pakistan, demanding $4.5 million in Bitcoin for decryption keys and to stop the release of stolen data. Other companies known to have suffered such attacks include the French system and software consultancy Sopra Steria; the Japanese game developer Capcom; the Italian liquor company Campari Group; the US military missile contractor Westech; the global aerospace and electronics engineering group ST Engineering; travel management giant CWT, who paid $4.5M in Bitcoin to the Ragnar Locker ransomware operators; business services giant Conduent; even soccer club Manchester United. Research shows that in Q3 2020, nearly half of all ransomware cases included the threat of releasing stolen data, and the average ransom payment was $233,817 – up 30% compared to Q2 2020. And that’s just the average ransom paid.
Manual deploys worked surprisingly well while we were getting our services up and running. More and more features were added to mix to interact not just with k8s but also other GCP services. To avoid dealing with raw YAML files directly, we moved our k8s configuration management to Jsonnet. Jsonnet allowed us to add templates for commonly used paradigms and reuse them in different deployments. At the same time, we kept adding more k8s clusters. We added more geographically distributed clusters to run the servers handling incoming data to decrease latency perceived by our ingestion API clients. Around the end of 2018, we started evaluating a European Data Residency product. That required us to deploy another full copy of all our services in two zones in the European Union. We were now up to 12 separate clusters, and many of them ran the same code and had similar configurations. While manual deploys worked fine when we ran code in just two zones, it quickly became infeasible to keep 12 separate clusters in sync manually. Across all our teams, we run more than 100 separate services and deployments.
Generally, physics and financial systems are not easily associated in people's minds. Yet, principles and techniques originating from physics can be very effective in describing the processes taking place on financial markets. Modeling financial systems as networks can greatly enhance our understanding of phenomena that are relevant not only to researchers in economics and other disciplines, but also to ordinary citizens, public agencies and governments. The theory of Complex Networks represents a powerful framework for studying how shocks propagate in financial systems, identifying early-warning signals of forthcoming crises, and reconstructing hidden linkages in interbank systems. ... Here is where network theory comes into play, by clarifying the interplay between the structure of the network, the heterogeneity of the individual characteristics of financial actors and the dynamics of risk propagation, in particular contagion, i.e. the domino effect by which the instability of some financial institutions can reverberate to other institutions to which they are connected. The associated risk is indeed "systemic", i.e. both produced and faced by the system as a whole, as in collective phenomena studied in physics.
The trend involves a complex blend of geopolitical and cybersecurity factors, but the underlying reasons for its recent explosion are simple. Ransomware attacks have gotten incredibly easy to execute, and payment methods are now much more friendly to criminals. Meanwhile, businesses are growing increasingly reliant on digital infrastructure and more willing to pay ransoms, thereby increasing the incentive to break in. As the New York Times notes, for years “criminals had to play psychological games to trick people into handing over bank passwords and have the technical know-how to siphon money out of secure personal accounts.” Now, young Russians with a criminal streak and a cash imbalance can simply buy the software and learn the basics on YouTube tutorials, or by getting help from syndicates like DarkSide — who even charge clients a fee to set them up to hack into businesses in exchange for a portion of the proceeds. The breach of the education publisher involving the false pedophile threat was a successful example of such a criminal exchange. Meanwhile, Bitcoin has made it much easier for cybercriminals to collect on their schemes.
Quote for the day:
"To make a decision, all you need is authority. To make a good decision, you also need knowledge, experience, and insight." -- Denise Moreland