Jonathan C. Roberts

Jonathan Roberts, Timo Lüddecke, Rehan Sheikh et al. · cambridge

Multimodal large language models (MLLMs) have shown remarkable capabilities across a broad range of tasks but their knowledge and abilities in the geographic and geospatial domains are yet to be explored, despite potential wide-ranging benefits to navigation, environmental research, urban development, and disaster response. We conduct a series of experiments exploring various vision capabilities of MLLMs within these domains, particularly focusing on the frontier model GPT-4V, and benchmark its performance against open-source counterparts. Our methodology involves challenging these models with a small-scale geographic benchmark consisting of a suite of visual tasks, testing their abilities across a spectrum of complexity. The analysis uncovers not only where such models excel, including instances where they outperform humans, but also where they falter, providing a balanced view of their capabilities in the geographic domain. To enable the comparison and evaluation of future models, our benchmark will be publicly released.

Jonathan Roberts, Kai Han, Samuel Albanie · cambridge

Interpreting remote sensing imagery enables numerous downstream applications ranging from land-use planning to deforestation monitoring. Robustly classifying this data is challenging due to the Earth's geographic diversity. While many distinct satellite and aerial image classification datasets exist, there is yet to be a benchmark curated that suitably covers this diversity. In this work, we introduce SATellite ImageNet (SATIN), a metadataset curated from 27 existing remotely sensed datasets, and comprehensively evaluate the zero-shot transfer classification capabilities of a broad range of vision-language (VL) models on SATIN. We find SATIN to be a challenging benchmark-the strongest method we evaluate achieves a classification accuracy of 52.0%. We provide a $\href{https://satinbenchmark.github.io}{\text{public leaderboard}}$ to guide and track the progress of VL models in this important domain.

Jonathan Roberts, Kai Han, Samuel Albanie · cambridge

As the context limits of Large Language Models (LLMs) increase, the range of possible applications and downstream functions broadens. In many real-world tasks, decisions depend on details scattered across collections of often disparate documents containing mostly irrelevant information. Long-context LLMs appear well-suited to this form of complex information retrieval and reasoning, which has traditionally proven costly and time-consuming. However, although the development of longer context models has seen rapid gains in recent years, our understanding of how effectively LLMs use their context has not kept pace. To address this, we conduct a set of retrieval experiments designed to evaluate the capabilities of 17 leading LLMs, such as their ability to follow threads of information through the context window. Strikingly, we find that many models are remarkably threadsafe: capable of simultaneously following multiple threads without significant loss in performance. Still, for many models, we find the effective context limit is significantly shorter than the supported context length, with accuracy decreasing as the context window grows. Our study also highlights the important point that token counts from different tokenizers should not be directly compared -- they often correspond to substantially different numbers of written characters. We release our code and long-context experimental data.

Jonathan Roberts, Kai Han, Neil Houlsby et al. · cambridge

Large multimodal models (LMMs) have proven flexible and generalisable across many tasks and fields. Although they have strong potential to aid scientific research, their capabilities in this domain are not well characterised. A key aspect of scientific research is the ability to understand and interpret figures, which serve as a rich, compressed source of complex information. In this work, we present SciFIBench, a scientific figure interpretation benchmark consisting of 2000 questions split between two tasks across 8 categories. The questions are curated from arXiv paper figures and captions, using adversarial filtering to find hard negatives and human verification for quality control. We evaluate 28 LMMs on SciFIBench, finding it to be a challenging benchmark. Finally, we investigate the alignment and reasoning faithfulness of the LMMs on augmented question sets from our benchmark. We release SciFIBench to encourage progress in this domain.

Jonathan Roberts, Mohammad Reza Taesiri, Ansh Sharma et al. · cambridge, oxford

Large Multimodal Models (LMMs) exhibit major shortfalls when interpreting images and, by some measures, have poorer spatial cognition than small children or animals. Despite this, they attain high scores on many popular visual benchmarks, with headroom rapidly eroded by an ongoing surge of model progress. To address this, there is a pressing need for difficult benchmarks that remain relevant for longer. We take this idea to its limit by introducing ZeroBench-a lightweight visual reasoning benchmark that is entirely impossible for contemporary frontier LMMs. Our benchmark consists of 100 manually curated questions and 334 less difficult subquestions. We evaluate 20 LMMs on ZeroBench, all of which score 0.0%, and rigorously analyse the errors. To encourage progress in visual understanding, we publicly release ZeroBench.

Wenye Lin, Jonathan Roberts, Yunhan Yang et al. · cambridge

Large Language Models (LLMs) are increasingly deployed in real-world applications that demand complex reasoning. To track progress, robust benchmarks are required to evaluate their capabilities beyond superficial pattern recognition. However, current LLM reasoning benchmarks often face challenges such as insufficient interpretability, performance saturation or data contamination. To address these challenges, we introduce GAMEBoT, a gaming arena designed for rigorous and transparent assessment of LLM reasoning capabilities. GAMEBoT decomposes complex reasoning in games into predefined modular subproblems. This decomposition allows us to design a suite of Chain-of-Thought (CoT) prompts that leverage domain knowledge to guide LLMs in addressing these subproblems before action selection. Furthermore, we develop a suite of rule-based algorithms to generate ground truth for these subproblems, enabling rigorous validation of the LLMs' intermediate reasoning steps. This approach facilitates evaluation of both the quality of final actions and the accuracy of the underlying reasoning process. GAMEBoT also naturally alleviates the risk of data contamination through dynamic games and head-to-head LLM competitions. We benchmark 17 prominent LLMs across eight games, encompassing various strategic abilities and game characteristics. Our results suggest that GAMEBoT presents a significant challenge, even when LLMs are provided with detailed CoT prompts. Project page: https://visual-ai.github.io/gamebot

Saiful Khan, Scott Jones, Benjamin Bach et al.

We present a method to create storytelling visualization with time series data. Many personal decisions nowadays rely on access to dynamic data regularly, as we have seen during the COVID-19 pandemic. It is thus desirable to construct storytelling visualization for dynamic data that is selected by an individual for a specific context. Because of the need to tell data-dependent stories, predefined storyboards based on known data cannot accommodate dynamic data easily nor scale up to many different individuals and contexts. Motivated initially by the need to communicate time series data during the COVID-19 pandemic, we developed a novel computer-assisted method for meta-authoring of stories, which enables the design of storyboards that include feature-action patterns in anticipation of potential features that may appear in dynamically arrived or selected data. In addition to meta-storyboards involving COVID-19 data, we also present storyboards for telling stories about progress in a machine learning workflow. Our approach is complementary to traditional methods for authoring storytelling visualization, and provides an efficient means to construct data-dependent storyboards for different data-streams of similar contexts.

Jonathan Roberts, Timo Lüddecke, Sowmen Das et al.

Large language models (LLMs) have shown remarkable capabilities across a broad range of tasks involving question answering and the generation of coherent text and code. Comprehensively understanding the strengths and weaknesses of LLMs is beneficial for safety, downstream applications and improving performance. In this work, we investigate the degree to which GPT-4 has acquired factual geographic knowledge and is capable of using this knowledge for interpretative reasoning, which is especially important for applications that involve geographic data, such as geospatial analysis, supply chain management, and disaster response. To this end, we design and conduct a series of diverse experiments, starting from factual tasks such as location, distance and elevation estimation to more complex questions such as generating country outlines and travel networks, route finding under constraints and supply chain analysis. We provide a broad characterisation of what GPT-4 (without plugins or Internet access) knows about the world, highlighting both potentially surprising capabilities but also limitations.

J. C. Roberts, J. W. Mearman, P. W. S. Butcher et al.

We believe that 3D visualisations should not be used alone; by coincidentally displaying alternative views the user can gain the best understanding of all situations. The different presentations signify manifold meanings and afford different tasks. Natural 3D worlds implicitly tell many stories. For instance, walking into a living room, seeing the TV, types of magazines, pictures on the wall, tells us much about the occupiers: their occupation, standards of living, taste in design, whether they have kids, and so on. How can we similarly create rich and diverse 3D visualisation presentations? How can we create visualisations that allow people to understand different stories from the data? In a multivariate 2D visualisation a developer may coordinate and link many views together to provide exploratory visualisation functionality. But how can this be achieved in 3D and in immersive visualisations? Different visualisation types, each have specific uses, and each has the potential to tell or evoke a different story. Through several use-cases, we discuss challenges of 3D visualisation, and present our argument for concurrent and coordinated visualisations of alternative styles, and encourage developers to consider using alternative representations with any 3D view, even if that view is displayed in a virtual, augmented or mixed reality setup.

Jonathan C. Roberts

We present how we have created a series of bilingual (English and Welsh) STEM activities focusing on rainbows, colours, light and optical effects. The activities were motivated by the many rainbows that appeared in windows in the UK, in support of the National Health Service at the start of the coronavirus pandemic. Rainbows are hopeful and are very fitting to be used as a positive iconic image at a time of much uncertainty. In this paper we explain how we have developed and organised the activities, focusing on colours, computer graphics and computer programming. Each lesson contains one or more activities, which enable people to take an active role in their learning. We have carefully prepared and organised several processes to guide academic colleagues to create and publish different activities in the theme. Which means that the activities appear similarly structured, can be categorised and searched in a consistent way. This structure can act as a blueprint for others to follow and apply to develop their own online course. The activities incrementally take people through learning about colour, rainbows, planning what to program, design and strategies to create colourful pictures using simple computer graphics principles based in processing.org.

Jonathan C. Roberts, Peter Butcher, Ann Sherlock et al.

Administrative justice concerns the relationships between individuals and the state. It includes redress and complaints on decisions of a child's education, social care, licensing, planning, environment, housing and homelessness. However, if someone has a complaint or an issue, it is challenging for people to understand different possible redress paths and explore what path is suitable for their situation. Explanatory visualisation has the potential to display these paths of redress in a clear way, such that people can see, understand and explore their options. The visualisation challenge is further complicated because information is spread across many documents, laws, guidance and policies and requires judicial interpretation. Consequently, there is not a single database of paths of redress. In this work we present how we have co-designed a system to visualise administrative justice paths of redress. Simultaneously, we classify, collate and organise the underpinning data, from expert workshops, heuristic evaluation and expert critical reflection. We make four contributions: (i) an application design study of the explanatory visualisation tool (Artemus), (ii) coordinated and co-design approach to aggregating the data, (iii) two in-depth case studies in housing and education demonstrating explanatory paths of redress in administrative law, and (iv) reflections on the expert co-design process and expert data gathering and explanatory visualisation for administrative justice and law.

Yaqub Jonmohamadi, Shahnewaz Ali, Fengbei Liu et al.

Minimally invasive surgery (MIS) has many documented advantages, but the surgeon's limited visual contact with the scene can be problematic. Hence, systems that can help surgeons navigate, such as a method that can produce a 3D semantic map, can compensate for the limitation above. In theory, we can borrow 3D semantic mapping techniques developed for robotics, but this requires finding solutions to the following challenges in MIS: 1) semantic segmentation, 2) depth estimation, and 3) pose estimation. In this paper, we propose the first 3D semantic mapping system from knee arthroscopy that solves the three challenges above. Using out-of-distribution non-human datasets, where pose could be labeled, we jointly train depth+pose estimators using selfsupervised and supervised losses. Using an in-distribution human knee dataset, we train a fully-supervised semantic segmentation system to label arthroscopic image pixels into femur, ACL, and meniscus. Taking testing images from human knees, we combine the results from these two systems to automatically create 3D semantic maps of the human knee. The result of this work opens the pathway to the generation of intraoperative 3D semantic mapping, registration with pre-operative data, and robotic-assisted arthroscopy

M. Chen, A. Abdul-Rahman, D. Archambault et al.

The effort for combating the COVID-19 pandemic around the world has resulted in a huge amount of data, e.g., from testing, contact tracing, modelling, treatment, vaccine trials, and more. In addition to numerous challenges in epidemiology, healthcare, biosciences, and social sciences, there has been an urgent need to develop and provide visualisation and visual analytics (VIS) capacities to support emergency responses under difficult operational conditions. In this paper, we report the experience of a group of VIS volunteers who have been working in a large research and development consortium and providing VIS support to various observational, analytical, model-developmental and disseminative tasks. In particular, we describe our approaches to the challenges that we have encountered in requirements analysis, data acquisition, visual design, software design, system development, team organisation, and resource planning. By reflecting on our experience, we propose a set of recommendations as the first step towards a methodology for developing and providing rapid VIS capacities to support emergency responses.

Xi Chen, Wei Zeng, Yanna Lin et al.

Multiple-view visualization (MV) is a layout design technique often employed to help users see a large number of data attributes and values in a single cohesive representation. Because of its generalizability, the MV design has been widely adopted by the visualization community to help users examine and interact with large, complex, and high-dimensional data. However, although ubiquitous, there has been little work to categorize and analyze MVs in order to better understand its design space. As a result, there has been little to no guideline in how to use the MV design effectively. In this paper, we present an in-depth study of how MVs are designed in practice. We focus on two fundamental measures of multiple-view patterns: composition, which quantifies what view types and how many are there; and configuration, which characterizes spatial arrangement of view layouts in the display space. We build a new dataset containing 360 images of MVs collected from IEEE VIS, EuroVis, and PacificVis publications 2011 to 2019, and make fine-grained annotations of view types and layouts for these visualization images. From this data we conduct composition and configuration analyses using quantitative metrics of term frequency and layout topology. We identify common practices around MVs, including relationship of view types, popular view layouts, and correlation between view types and layouts. We combine the findings into a MV recommendation system, providing interactive tools to explore the design space, and support example-based design.