We study personalized figure caption generation using author profile data from scientific papers. Our experiments demonstrate that rich author profile data, combined with relevant metadata, can significantly improve the personalization performance of multimodal large language models. However, we also reveal a fundamental trade-off between matching author style and maintaining caption quality. Our findings offer valuable insights and future directions for developing practical caption automation systems that balance both objectives. This work was conducted as part of the 3rd SciCap challenge.

We propose a novel approach to effectively detect cloned identities of social-sensor cloud service providers (i.e. social media users) in the face of incomplete non-privacy-sensitive profile data. Named ICD-IPD, the proposed approach first extracts account pairs with similar usernames or screen names from a given set of user accounts collected from a social media. It then learns a multi-view representation associated with a given account and extracts two categories of features for every single account. These two categories of features include profile and Weighted Generalised Canonical Correlation Analysis (WGCCA)-based features that may potentially contain missing values. To counter the impact of such missing values, a missing value imputer will next impute the missing values of the aforementioned profile and WGCCA-based features. After that, the proposed approach further extracts two categories of augmented features for each account pair identified previously, namely, 1) similarity and 2) differences-based features. Finally, these features are concatenated and fed into a Light Gradient Boosting Machine classifier to detect identity cloning. We evaluated and compared the proposed approach against the existing state-of-the-art identity cloning approaches and other machine or deep learning models atop a real-world dataset. The experimental results show that the proposed approach outperforms the state-of-the-art approaches and models in terms of Precision, Recall and F1-score.

In this study, we critically examine the foundational premise of algorithmic recourse - a process of generating counterfactual action plans (i.e., recourses) assisting individuals to reverse adverse decisions made by AI systems. The assumption underlying algorithmic recourse is that individuals accept and act on recourses that minimize the gap between their current and desired states. This assumption, however, remains empirically unverified. To address this issue, we conducted a user study with 362 participants and assessed whether minimizing the distance function, a metric of the gap between the current and desired states, indeed prompts them to accept and act upon suggested recourses. Our findings reveal a nuanced landscape: participants' acceptance of recourses did not correlate with the recourse distance. Moreover, participants' willingness to act upon recourses peaked at the minimal recourse distance but was otherwise constant. These findings cast doubt on the prevailing assumption of algorithmic recourse research and signal the need to rethink the evaluation functions to pave the way for human-centered recourse generation.

Large Language Models (LLMs) deployed on edge devices learn through fine-tuning and updating a certain portion of their parameters. Although such learning methods can be optimized to reduce resource utilization, the overall required resources remain a heavy burden on edge devices. Instead, Retrieval-Augmented Generation (RAG), a resource-efficient LLM learning method, can improve the quality of the LLM-generated content without updating model parameters. However, the RAG-based LLM may involve repetitive searches on the profile data in every user-LLM interaction. This search can lead to significant latency along with the accumulation of user data. Conventional efforts to decrease latency result in restricting the size of saved user data, thus reducing the scalability of RAG as user data continuously grows. It remains an open question: how to free RAG from the constraints of latency and scalability on edge devices? In this paper, we propose a novel framework to accelerate RAG via Computing-in-Memory (CiM) architectures. It accelerates matrix multiplications by performing in-situ computation inside the memory while avoiding the expensive data transfer between the computing unit and memory. Our framework, Robust CiM-backed RAG (RoCR), utilizing a novel contrastive learning-based training method and noise-aware training, can enable RAG to efficiently search profile data with CiM. To the best of our knowledge, this is the first work utilizing CiM to accelerate RAG.

There is an interesting vulnerability in video deepfakes that, to date, has been generally overlooked by the security research community, perhaps because'live', real-time deepfakes in video calls have not been a major cause for concern until very recently. For a number of reasons, which we'll examine in this article, deepfakes are not usually very good at recreating profile views: The above examples are taken* from a session with tech exponent and commenter Bob Doyle, who agreed to run some tests with us, using DeepFaceLive to change his appearance to that of a series of popular celebrities. DeepFaceLive is a live-streaming version of the popular DeepFaceLab software, and is capable of creating alternate video identities in real-time. For these tests, Bob used a mix of models downloaded from deepfake communities, as well as some community-trained models that come bundled with the DeepFaceLive application. From more or less face-on viewpoints, most of the celebrity recreations are quite effective, and some are very convincing even at fairly acute angles – until the facial angle hits a full 90 .

A recent study has used machine learning analysis techniques to chart the readability, usefulness, length and complexity of more than 50,000 privacy policies on popular websites in a period covering 25 years from 1996 to 2021. The research concludes that the average reader would need to devote 400 hours of'annual reading time' (more than an hour a day) in order to penetrate the growing word counts, obfuscating language and vague language use that characterize the modern privacy policies of some of the most-frequented websites. 'The average policy length has almost doubled in the last ten years, with 2159 words in March 2011 and 4191 words in March 2021, and almost quadrupled since 2000 (1146 words).' The mean word count and sentence count among the corpus studied, over a 25 year period. Though the rate of increase in length spiked when the GDPR and the California Consumer Privacy Act (CCPA) protections came into force, the paper discounts these variations as'small effect sizes' which appear to be insignificant against the broader long-term trend.

This is a guest post by the Jira Cloud Performance Team at Atlassian. In their own words, Atlassian's mission is to unleash the potential in every team. Our products help teams organize, discuss, and complete their work. And what teams do can change the world. We have helped NASA teams design the Mars Rover, Cochlear teams develop hearing implants and hundreds of thousands of other teams do amazing things.

Advertisers know ads are most successful when they promote the right product at the right time to the right customer. However, getting the combination of product, timing, customer and channel correct is incredibly difficult. Advertisers have long sought after the goal of hyperpersonalization, where individual promotions can be tailored and targeted to individual people at the right time, in the right format and through the right channel that will meet an immediate need and result in a greater chance of conversion. With traditional approaches to advertising -- print ads, TV commercials or billboards -- it's impossible to create highly personalized ads, as the audience is bound to be relatively diverse. The advent of IoT and mobile technology, combined with big data gleaned from customer interactions, gives advertisers the opportunity to target their customers through incredibly personal and timely information about customer behaviors.

Forward by the team at Go Wine: This interview with John Seaton at Cendyn shows how Customer Relationship Management in 2018, is being impacted by artificial intelligence, globalization travel trends and more. In your opinion, what are the top three trends that hoteliers should be aware of going into 2018? A. The evolving nature of the guest experience and keeping up with guests' needs and expectations is a huge focus for the hospitality industry. For a hotel, managing the customer relationship is one of the most critical elements of gaining and increasing loyalty, and yet can be the most difficult for hotels to master, as customers interact with them via a burgeoning number of contact points: email, mobile, social media, at the front desk and throughout the hotel property. Never before has technology played a more important role in improving what is ultimately the human experience of hospitality, both in terms of curating and providing it, but also in the way that customers express their gratitude for that experience in the form of loyalty. B. Secondly, understanding the capabilities of artificial intelligence (AI) and how that can focus and positively affect the interaction between the guest experience and the hotel.

Google has previously taught its artificial intelligence to play games, and it's even capable of creating its own encryption. Now, its language translation tool has used machine learning to create a'language' all of its own. In September, the search giant turned on its Google Neural Machine Translation (GNMT) system to help it automatically improve how it translates languages. The machine learning system analyses and makes sense of languages by looking at entire sentences – rather than individual phrases or words. Following several months of testing, the researchers behind the AI have seen it be able to blindly translate languages even if it's never studied one of the languages involved in the translation.