Attribution methods, which employ heatmaps to identify the most influential regions of an image that impact model decisions, have gained widespread popularity as a type of explainability method. However, recent research has exposed the limited practical value of these methods, attributed in part to their narrow focus on the most prominent regions of an image -- revealing "where" the model looks, but failing to elucidate "what" the model sees in those areas. In this work, we try to fill in this gap with CRAFT -- a novel approach to identify both "what" and "where" by generating concept-based explanations. We introduce 3 new ingredients to the automatic concept extraction literature: (i) a recursive strategy to detect and decompose concepts across layers, (ii) a novel method for a more faithful estimation of concept importance using Sobol indices, and (iii) the use of implicit differentiation to unlock Concept Attribution Maps. We conduct both human and computer vision experiments to demonstrate the benefits of the proposed approach. We show that the proposed concept importance estimation technique is more faithful to the model than previous methods. When evaluating the usefulness of the method for human experimenters on a human-centered utility benchmark, we find that our approach significantly improves on two of the three test scenarios. Our code is freely available at github.com/deel-ai/Craft.

Over the past decade, programmatic advertising has received a great deal of attention in the online advertising industry. A real-time bidding (RTB) system is rapidly becoming the most popular method to buy and sell online advertising impressions. Within the RTB system, demand-side platforms (DSP) aim to spend advertisers' campaign budgets efficiently while maximizing profit, seeking impressions that result in high user responses, such as clicks or installs. In the current study, we investigate the process of predicting a mobile gaming app installation from the point of view of a particular DSP, while paying attention to user privacy, and exploring the trade-off between privacy preservation and model performance. There are multiple levels of potential threats to user privacy, depending on the privacy leaks associated with the data-sharing process, such as data transformation or de-anonymization. To address these concerns, privacy-preserving techniques were proposed, such as cryptographic approaches, for training privacy-aware machine-learning models. However, the ability to train a mobile gaming app installation prediction model without using user-level data, can prevent these threats and protect the users' privacy, even though the model's ability to predict may be impaired. Additionally, current laws might force companies to declare that they are collecting data, and might even give the user the option to opt out of such data collection, which might threaten companies' business models in digital advertising, which are dependent on the collection and use of user-level data. We conclude that privacy-aware models might still preserve significant capabilities, enabling companies to make better decisions, dependent on the privacy-efficacy trade-off utility function of each case.

Successful analytics solutions that provide valuable insights often hinge on the connection of various data sources. While it is often feasible to generate larger data pools within organizations, the application of analytics within (inter-organizational) business networks is still severely constrained. As data is distributed across several legal units, potentially even across countries, the fear of disclosing sensitive information as well as the sheer volume of the data that would need to be exchanged are key inhibitors for the creation of effective system-wide solutions -- all while still reaching superior prediction performance. In this work, we propose a meta machine learning method that deals with these obstacles to enable comprehensive analyses within a business network. We follow a design science research approach and evaluate our method with respect to feasibility and performance in an industrial use case. First, we show that it is feasible to perform network-wide analyses that preserve data confidentiality as well as limit data transfer volume. Second, we demonstrate that our method outperforms a conventional isolated analysis and even gets close to a (hypothetical) scenario where all data could be shared within the network. Thus, we provide a fundamental contribution for making business networks more effective, as we remove a key obstacle to tap the huge potential of learning from data that is scattered throughout the network.

Motivated by multi-center biomedical studies that cannot share individual data due to privacy and ownership concerns, we develop communication-efficient iterative distributed algorithms for estimation and inference in the high-dimensional sparse Cox proportional hazards model. We demonstrate that our estimator, even with a relatively small number of iterations, achieves the same convergence rate as the ideal full-sample estimator under very mild conditions. To construct confidence intervals for linear combinations of high-dimensional hazard regression coefficients, we introduce a novel debiased method, establish central limit theorems, and provide consistent variance estimators that yield asymptotically valid distributed confidence intervals. In addition, we provide valid and powerful distributed hypothesis tests for any coordinate element based on a decorrelated score test. We allow time-dependent covariates as well as censored survival times. Extensive numerical experiments on both simulated and real data lend further support to our theory and demonstrate that our communication-efficient distributed estimators, confidence intervals, and hypothesis tests improve upon alternative methods.

The US Federal Trade Commission is paying close attention to developments in artificial intelligence to ensure the field isn't dominated by the major tech platforms, Chair Lina Khan said Monday. "As you have machine learning that depends on huge amounts of data and also a huge amount of storage, we need to be very vigilant to make sure that this is not just another site for big companies to become bigger," Khan said at an event hosted by the Justice Department in Washington.

The AI era is upon us, with what seems like new advances every week, pushing the technology to new heights. Between Google, OpenAI, Microsoft and a raft of other companies, new developments that can ease the way we live and work are accessible to people more than ever before. It's little wonder, then, that businesses are starting to consider how best to integrate AI into their processes to reap the benefits. But thinking before acting is vital in such a fast-moving space. The first-mover advantage that businesses seek out can quickly be negated by the regulatory risks of irresponsible use of AI. "Lots of companies talk about AI, but only a few of them can talk about responsible AI," says Vikash Khatri, senior vice-president for artificial intelligence at Afiniti, which provides AI that pairs customers and contact-centre agents based on how well they are likely to interact. "Yet, it's vital that responsibility be front of mind when considering any deployment of AI – the risks of not considering that are too great."

In the popular imagination, AI is often portrayed as a Frankenstein-esque mistake: a creation that comes to destroy its creator. Stephen Hawking was once asked, "What could a robot do that I couldn't then fight back with by just unplugging him?" There's a story that scientists built an intelligent computer. The first question they asked it was: "Is there a God?" The computer replied: "There is now."

The internet has been enthralled with the rise of AI starting with the introduction of the Artificial Intelligence image generation program DALL-E 2 to the public in November 2022. Programs like ChatGPT have already begun to change how we use the internet, and huge organizations such as Microsoft and the British government have started investing hundreds of millions of dollars into creating new AI technology. AI is going to be around for a while and, like with every other advancement in technology, will inevitably cause some problems. Programs like ChatGPT have the potential to radically change how we interact with our devices, but that potential needs to be realized while understanding the limits of the technology and protecting the rights of workers. Before we unpack what limits may need to be placed on AI, it is important to understand what makes the tech useful. Recently, the introduction of the massively improved GPT-4 has shown that AI has the potential to be a great asset to society.

She wanted to know if I had any suggestions, and asked what I thought all the new advances meant for lawmakers. I've spent a few days thinking, reading, and chatting with the experts about this, and my answer morphed into this newsletter. Though GPT-4 is the standard bearer, it's just one of many high-profile generative AI releases in the past few months: Google, Nvidia, Adobe, and Baidu have all announced their own projects. In short, generative AI is the thing that everyone is talking about. And though the tech is not new, its policy implications are months if not years from being understood.

A.I. image generators are divisive. But few can deny that they have gotten really good. Within seconds, you can type in a prompt to make a photorealistic image of Donald Trump getting arrested or turn your strangest idea into something tangible. Over the coming years, A.I. companies will release even more advanced models that will remind us that this is just the beginning. At least one of these tools will be different in an important way: It will be prohibited from seeing 80 million of the images that helped teach its predecessors to draw and paint.