Visual encoders have become fundamental components in modern computer vision pipelines. However, ensuring robustness against adversarial perturbations remains a critical challenge. Recent efforts have explored both supervised and unsupervised adversarial fine-tuning strategies. We identify two key limitations in these approaches: (i) they often suffer from instability, especially during the early stages of fine-tuning, resulting in suboptimal convergence and degraded performance on clean data, and (ii) they exhibit a suboptimal trade-off between robustness and clean data accuracy, hindering the simultaneous optimization of both objectives. To overcome these challenges, we propose Lagrangian-Optimized Robust Embeddings (LORE), a novel unsupervised adversarial fine-tuning framework. LORE utilizes constrained optimization, which offers a principled approach to balancing competing goals, such as improving robustness while preserving nominal performance. By enforcing embedding-space proximity constraints, LORE effectively maintains clean data performance throughout adversarial fine-tuning. Extensive experiments show that LORE significantly improves zero-shot adversarial robustness with minimal degradation in clean data accuracy. Furthermore, we demonstrate the effectiveness of the adversarially fine-tuned CLIP image encoder in out-of-distribution generalization and enhancing the interpretability of image embeddings.

HP has announced a lower-than-expected profit outlook for the coming year. HP has announced a lower-than-expected profit outlook for the coming year. Up to 6,000 jobs are to go at HP worldwide in the next three years as the US computer and printer maker increasingly adopts AI to speed up product development. Announcing a lower-than-expected profit outlook for the coming year, HP said it would cut between 4,000 and 6,000 jobs by the end of October 2028. It has about 56,000 employees.

Personalizing large language models (LLMs) to accommodate diverse user preferences is essential for enhancing alignment and user satisfaction. Traditional reinforcement learning from human feedback (RLHF) approaches often rely on monolithic value representations, limiting their ability to adapt to individual preferences. We introduce a novel framework that leverages low-rank preference modeling to efficiently learn and generalize user-specific reward functions. By representing reward functions in a low-dimensional subspace and modeling individual preferences as weighted combinations of shared basis functions, our approach avoids rigid user categorization while enabling scalability and few-shot adaptation. We validate our method on multiple preference datasets, demonstrating superior generalization to unseen users and improved accuracy in preference prediction tasks.

This week I took my son, Zac, to see the new Minecraft movie, which is hardly a remarkable statement in the highly video game-branded world of 21st-century cinema – except that what followed was not typical at all. As you may have seen from a number of bewildered news reports over the last few days, A Minecraft Movie has quickly engendered a community of, let's say, highly engaged and enthusiastic fans. Spurred on by TikTok meme posts, vast portions of the film's audience are now yelling out key lines of dialogue as they happen and singing along to the songs. In one key moment where a rare character from the game – the zombie chicken jockey – is introduced, they go absolutely crazy, throwing drinks and popcorn around, and in some US cinemas, getting escorted from the screening by police. The reaction was a little more muted in our tiny independent cinema in Frome, but still, there were rows of teenagers who had clearly seen all the TikTok posts telling them which lines to shout along to, and went to throw stuff, and they were extremely excited to be doing so, a few surreptitiously filming their mates' reactions so they could add to the social media carnage.

Modern conversational agents, including anime-themed chatbots, are frequently reactive and personality-driven but fail to capture the dynamic nature of human interactions. We propose an event-driven dialogue framework to address these limitations by embedding dynamic events in conversation prompts and fine-tuning models on character-specific data. Evaluations on GPT-4 and comparisons with industry-leading baselines demonstrate that event-driven prompts significantly improve conversational engagement and naturalness while reducing hallucinations. This paper explores the application of this approach in creating lifelike chatbot interactions within the context of Honkai: Star Rail, showcasing the potential for dynamic event-based systems to transform role-playing and interactive dialogue.

Retrieval-based question answering systems often suffer from positional bias, leading to suboptimal answer generation. We propose LoRE (Logit-Ranked Retriever Ensemble), a novel approach that improves answer accuracy and relevance by mitigating positional bias. LoRE employs an ensemble of diverse retrievers, such as BM25 and sentence transformers with FAISS indexing. A key innovation is a logit-based answer ranking algorithm that combines the logit scores from a large language model (LLM), with the retrieval ranks of the passages. Experimental results on NarrativeQA, SQuAD demonstrate that LoRE significantly outperforms existing retrieval-based methods in terms of exact match and F1 scores. On SQuAD, LoRE achieves 14.5\%, 22.83\%, and 14.95\% improvements over the baselines for ROUGE-L, EM, and F1, respectively. Qualitatively, LoRE generates more relevant and accurate answers, especially for complex queries.

The Steam Summer Sale is in full swing, and there are deals to be had everywhere you look. But most of them--especially the more popular and more recent games--only have modest discounts at best. I've picked out some of my favorites that have the biggest savings. In no particular order, here are the best Steam Summer Sale deals on the games I personally love and think everyone should check out. Fallout is all the rage at the moment thanks to the Amazon series, but sadly it's been years since the last mainline game--and it'll be years more before a new one comes out.

Channelling the sci-fi military satire and extreme gloopy gore of Starship Troopers, Helldivers 2 was a surprise mega hit on its launch in February. Looking back we shouldn't have been shocked: it delivers engrossing, hilarious co-op action in a range of desolate landscapes against horrible insects and crazed robots, and it makes each fight feel like part of a much wider story – a factor heightened by Arrowhead Game Studios' excellent use of social media channels. What we said: "Everything about this game is ridiculous, including how good it is at what it sets out to do." Read the full review The set-up is not wildly promising: you play as a blob, stuck down a well. But this winding puzzle platformer is an ethereal delight, dense in atmosphere, with visuals that look like a lost 1980s arcade game glimpsed through a window. You explore, you learn new skills, you see lovely creatures.

Probabilistic classifiers output confidence scores along with their predictions, and these confidence scores should be calibrated, i.e., they should reflect the reliability of the prediction. Confidence scores that minimize standard metrics such as the expected calibration error (ECE) accurately measure the reliability on average across the entire population. However, it is in general impossible to measure the reliability of an individual prediction. In this work, we propose the local calibration error (LCE) to span the gap between average and individual reliability. For each individual prediction, the LCE measures the average reliability of a set of similar predictions, where similarity is quantified by a kernel function on a pretrained feature space and by a binning scheme over predicted model confidences. We show theoretically that the LCE can be estimated sample-efficiently from data, and empirically find that it reveals miscalibration modes that are more fine-grained than the ECE can detect. Our key result is a novel local recalibration method LoRe, to improve confidence scores for individual predictions and decrease the LCE. Experimentally, we show that our recalibration method produces more accurate confidence scores, which improves downstream fairness and decision making on classification tasks with both image and tabular data.

Instacart provides grocery delivery and pickup services in the US and Canada. Customers can use the service to order goods from participating stores, and a personal shopper will conduct the shopping for them. As one can imagine, the Instacart experience relies heavily on machine learning. Nearly every product and business innovation at the company is based on machine learning (ML), including helping users locate the ideal things in a catalog of more than 1 billion products and enabling 5,000 brand partners to connect their products to potential customers. In 2016, the company began creating its machine learning framework called Lore.