There has been an increasing awareness in the developer community, enterprises, and governments of software supply chain risks. Remediation efforts for vulnerabilities like Log4j and Spring4shell, and a 650% year-over-year increase in cyberattacks aimed at open source suppliers, have sharpened focus on the critical task of bolstering the security of open source software. Governments and regulators have taken notice and action, including the White House’s Executive Order 14028 on Improving the Nation’s Cybersecurity, followed by other governments and agencies around the world asserting new requirements and standards specifically focused on the software development lifecycle and the software supply chain.
…To further our commitment to help organizations strengthen their OSS software supply chain, we are announcing today a new Google Cloud product: our Assured Open Source Software service. Assured OSS enables enterprise and public sector users of open source software to easily incorporate the same OSS packages that Google uses into their own developer workflows. Read More
Daily Archives: June 2, 2022
VOS: Learning what you don’t know by virtual outlier synthesis
Out-of-distribution (OOD) detection has received much attention lately due to its importance in the safe deployment of neural networks. One of the key challenges is that models lack supervision signals from unknown data, and as a result, can produce overconfident predictions on OOD data. Previous approaches rely on real outlier datasets for model regularization, which can be costly and sometimes infeasible to obtain in practice. In this paper, we present VOS, a novel framework for OOD detection by adaptively synthesizing virtual outliers that can meaningfully regularize the model’s decision boundary during training. Specifically, VOS samples virtual outliers from the low-likelihood region of the class-conditional distribution estimated in the feature space. Alongside, we introduce a novel unknown aware training objective, which contrastively shapes the uncertainty space between the ID data and synthesized outlier data. VOS achieves competitive performance on both object detection and image classification models, reducing the FPR95 by up to 9.36% compared to the previous best method on object detectors. Read More
The Future of the Web Is Marketing Copy Generated by Algorithms
AS WE MOVE further into the 21st century, more and more aspects of our lives are being controlled by algorithms. Facebook decides which posts we see in our newsfeed, Google shows us the results of our searches based on their complex ranking system, and Amazon recommends products based on our past purchase history. It’s no wonder then that online marketing is becoming increasingly reliant on algorithms to create effective copywriting. So what does the future hold for the web—will marketing be dominated by machines, or will human creativity always be necessary? Read on to find out …
No human wrote that intro. It was generated by software from the copywriting service Jasper, inspired by the headline on this article. The first suggestion was too brief and lacked detail. The second, reproduced verbatim above, caused an editor to exclaim that she had received worse copy from professional writers.
Jasper can also generate content tuned for Facebook ads, marketing emails, and product descriptions. It’s part of a raft of startups that have adapted a text-generation technology known as GPT-3, from the artificial-intelligence company OpenAI, to feed one of the internet’s oldest urges—to create marketing copy that wins clicks and ranks highly on Google. Read More
Xanadu Launches Borealis, a 216 Squeezed-state Quibit Photonic Processor
Xanadu, a Toronto based quantum computing startup, has released their Borealis processor and has made it available on the Xanadu Cloud. They will also make it available soon on the Amazon Braket cloud service. In addition, Xanadu has demonstrated Quantum Supremacy with this device by running a Random Number Sampling experiment on it use Gaussian Boson Sampling. This demonstration is similar to those performed by Google in October 2019 with their superconducting based device and also by the University of Science and Technology China (USTC) with their Gaussian Boson Sampler in December of 2020. Read More
Google quietly bans deepfake training projects on Colab
Google has quietly banned deepfake projects on its Colaboratory (Colab) service, putting an end to the large-scale utilization of the platform’s resources for this purpose.
Colab is an online computing resource that allows researchers to run Python code directly through the browser while using free computing resources, including GPUs, to power their projects.
… Based on archive.org historical data, the ban took place earlier this month, with Google Research quietly adding deep fakes to the list of disallowed projects. Read More
Rick's Cafe AI 