We propose an enhanced zeroth-order stochastic Frank-Wolfe framework to address constrained finite-sum optimization problems, a structure prevalent in large-scale machine-learning applications. Our method introduces a novel double variance reduction framework that effectively reduces the gradient approximation variance induced by zeroth-order oracles and the stochastic sampling variance from finite-sum objectives. By leveraging this framework, our algorithm achieves significant improvements in query efficiency, making it particularly well-suited for high-dimensional optimization tasks. Specifically, for convex objectives, the algorithm achieves a query complexity of O(d \sqrt{n}/\epsilon ) to find an epsilon-suboptimal solution, where d is the dimensionality and n is the number of functions in the finite-sum objective. For non-convex objectives, it achieves a query complexity of O(d^{3/2}\sqrt{n}/\epsilon^2 ) without requiring the computation ofd partial derivatives at each iteration. These complexities are the best known among zeroth-order stochastic Frank-Wolfe algorithms that avoid explicit gradient calculations. Empirical experiments on convex and non-convex machine learning tasks, including sparse logistic regression, robust classification, and adversarial attacks on deep networks, validate the computational efficiency and scalability of our approach. Our algorithm demonstrates superior performance in both convergence rate and query complexity compared to existing methods.

INTRODUCTION Advances in vitreoretinal surgery have enabled interventions involving precisions previously deemed infeasible [1], with certified systems appearing on the market, e.g. the Preceyes Surgical System offering 20μm accuracy [2]. Despite their benefits, systems add a delay to the interventional workflow, which may be hindering their widespread adoption. A source of delay is the time required to get the system's micro-precise tool into the eye via the Trocar Entry Point (TEP). We will compare 3 approaches that use a teleoperation of the tool's position and orientation via the combination of co-manipulation and teleoperation. The The goal is to place a 0.5mm stainless steel rod within a task is complete when the participant deems the docking 1 mm custom trocar inserted into the inferior position of sufficient and extrudes the rod into the phantom via the a Bioniko Fundus Advanced Eye Phantom [4].

Automatic radar signal recognition (RSR) plays a pivotal role in electronic warfare (EW), as accurately classifying radar signals is critical for informing decision-making processes. Recent advances in deep learning have shown significant potential in improving RSR performance in domains with ample annotated data. However, these methods fall short in EW scenarios where annotated RF data are scarce or impractical to obtain. To address these challenges, we introduce a self-supervised learning (SSL) method which utilises masked signal modelling and RF domain adaption to enhance RSR performance in environments with limited RF samples and labels. Specifically, we investigate pre-training masked autoencoders (MAE) on baseband in-phase and quadrature (I/Q) signals from various RF domains and subsequently transfer the learned representation to the radar domain, where annotated data are limited. Empirical results show that our lightweight self-supervised ResNet model with domain adaptation achieves up to a 17.5% improvement in 1-shot classification accuracy when pre-trained on in-domain signals (i.e., radar signals) and up to a 16.31% improvement when pre-trained on out-of-domain signals (i.e., comm signals), compared to its baseline without SSL. We also provide reference results for several MAE designs and pre-training strategies, establishing a new benchmark for few-shot radar signal classification.

The increasing demand for robust security solutions across various industries has made Video Anomaly Detection (VAD) a critical task in applications such as intelligent surveillance, evidence investigation, and violence detection. Traditional approaches to VAD often rely on finetuning large pre-trained models, which can be computationally expensive and impractical for real-time or resource-constrained environments. To address this, MissionGNN introduced a more efficient method by training a graph neural network (GNN) using a fixed knowledge graph (KG) derived from large language models (LLMs) like GPT-4. While this approach demonstrated significant efficiency in computational power and memory, it faces limitations in dynamic environments where frequent updates to the KG are necessary due to evolving behavior trends and shifting data patterns. These updates typically require cloud-based computation, posing challenges for edge computing applications. In this paper, we propose a novel framework that facilitates continuous KG adaptation directly on edge devices, overcoming the limitations of cloud dependency. Our method dynamically modifies the KG through a three-phase process: pruning, alternating, and creating nodes, enabling real-time adaptation to changing data trends. This continuous learning approach enhances the robustness of anomaly detection models, making them more suitable for deployment in dynamic and resource-constrained environments.

Learning the intents of an agent, defined by its goals or motion style, is often extremely challenging from just a few examples. We refer to this problem as task concept learning and present our approach, Few-Shot Task Learning through Inverse Generative Modeling (FTL-IGM), which learns new task concepts by leveraging invertible neural generative models. The core idea is to pretrain a generative model on a set of basic concepts and their demonstrations. Then, given a few demonstrations of a new concept (such as a new goal or a new action), our method learns the underlying concepts through backpropagation without updating the model weights, thanks to the invertibility of the generative model. We evaluate our method in five domains -- object rearrangement, goal-oriented navigation, motion caption of human actions, autonomous driving, and real-world table-top manipulation. Our experimental results demonstrate that via the pretrained generative model, we successfully learn novel concepts and generate agent plans or motion corresponding to these concepts in (1) unseen environments and (2) in composition with training concepts.

When building a predictive model, it is often difficult to ensure that application-specific requirements are encoded by the model that will eventually be deployed. Consider researchers working on hate speech detection. They will have an idea of what is considered hate speech, but building a model that reflects their view accurately requires preserving those ideals throughout the workflow of data set construction and model training. Complications such as sampling bias, annotation bias, and model misspecification almost always arise, possibly resulting in a gap between the application specification and the model's actual behavior upon deployment. To address this issue for hate speech detection, we propose DefVerify: a 3-step procedure that (i) encodes a user-specified definition of hate speech, (ii) quantifies to what extent the model reflects the intended definition, and (iii) tries to identify the point of failure in the workflow. We use DefVerify to find gaps between definition and model behavior when applied to six popular hate speech benchmark datasets.

Artificial intelligence will be "mainlined into the veins" of the nation, ministers have announced, with a multibillion-pound investment in the UK's computing capacity despite widespread public fear about the technology's effects. Keir Starmer will launch a sweeping action plan to increase 20-fold the amount of AI computing power under public control by 2030 and deploy AI for everything from spotting potholes to freeing up teachers to teach. Labour's plan to "unleash" AI includes a personal pledge from the prime minister to make Britain "the world leader" in a sector that has been transformed by a series of significant breakthroughs in the last three years. The government plan features a potentially controversial scheme to unlock public data to help fuel the growth of AI businesses. This includes anonymised NHS data, which will be available for "researchers and innovators" to train their AI models.

The spectre of last week's bond market sell-off hangs over the government's artificial intelligence strategy. Investors, and voters, want to know where the growth is in the UK economy. Keir Starmer and the chancellor, Rachel Reeves, believes AI is a significant part of the answer. The UK has considerable strengths in AI, which can be loosely defined as computer systems performing tasks that typically require human intelligence (ranging from summarising a document to assessing a medical patient's symptoms and writing emails). Those strengths include the high-quality research and engineering talent coming out of UK universities and the fact that the country already hosts a number of leading AI companies, led by the UK-founded Google DeepMind.

Tehran, Iran – Iran's Foreign Ministry and judiciary have confirmed that Iranian national Mohammad Abedini, who was arrested in Italy at the behest of the United States, has been released. Abedini was returned to Tehran after being arrested as part of a "misunderstanding", Mizan, the official news outlet of the judiciary, said on Sunday. The report, also aired by state television, said his release was secured after talks between the Iranian intelligence ministry and the Italian intelligence service. Foreign Ministry spokesman Esmaeil Baghaei in a short statement welcomed the release of the Iranian national, who is accused by Washington of involvement with a January 2024 drone attack on a US outpost in Jordan that killed three American soldiers. He stressed the ministry would defend the rights of Iranian nationals abroad.

Russia claims it has captured two villages in eastern Ukraine where its forces have been steadily advancing for months, as Ukraine's president urged allies to deliver all the weapons they have promised to send to Kyiv. The Russian Defence Ministry said on Sunday that soldiers have captured the village of Yantarne in the eastern Donetsk region, about 10km (six miles) southwest of Kurakhove, a key logistics hub that Moscow claimed to have seized last week – a day after Russia's army said it had also taken new territory northwest of Kurakhove. The Defence Ministry added that soldiers had also captured the village of Kalinove in the northeastern Kharkiv region. The village is on the western bank of the Oskil River, which for a long time formed the front line between the two armies in the region. A Ukrainian official, quoted by the AFP news agency, said on Thursday that Russian forces had managed to establish a bridgehead on the western bank after crossing the river.