Gary Koplik

Gary Koplik, Nathan Borggren, Sam Voisin et al.

As Internet of Things (IoT) devices become both cheaper and more powerful, researchers are increasingly finding solutions to their scientific curiosities both financially and computationally feasible. When operating with restricted power or communications budgets, however, devices can only send highly-compressed data. Such circumstances are common for devices placed away from electric grids that can only communicate via satellite, a situation particularly plausible for environmental sensor networks. These restrictions can be further complicated by potential variability in the communications budget, for example a solar-powered device needing to expend less energy when transmitting data on a cloudy day. We propose a novel, topology-based, lossy compression method well-equipped for these restrictive yet variable circumstances. This technique, Topological Signal Compression, allows sending compressed signals that utilize the entirety of a variable communications budget. To demonstrate our algorithm's capabilities, we perform entropy calculations as well as a classification exercise on increasingly topologically simplified signals from the Free-Spoken Digit Dataset and explore the stability of the resulting performance against common baselines.

Kenneth Ball, Erin Taylor, Nirav Patel et al.

Advances in machine learning, especially the introduction of transformer architectures and vision transformers, have led to the development of highly capable computer vision foundation models. The segment anything model (known colloquially as SAM and more recently SAM 2), is a highly capable foundation model for segmentation of natural images and has been further applied to medical and scientific image segmentation tasks. SAM relies on prompts -- points or regions of interest in an image -- to generate associated segmentations. In this manuscript we propose the use of a geometrically motivated prompt generator to produce prompt points that are colocated with particular features of interest. Focused prompting enables the automatic generation of sensitive and specific segmentations in a scientific image analysis task using SAM with relatively few point prompts. The image analysis task examined is the segmentation of plant roots in rhizotron or minirhizotron images, which has historically been a difficult task to automate. Hand annotation of rhizotron images is laborious and often subjective; SAM, initialized with GeomPrompt local ridge prompts has the potential to dramatically improve rhizotron image processing. The authors have concurrently released an open source software suite called geomprompt https://pypi.org/project/geomprompt/ that can produce point prompts in a format that enables direct integration with the segment-anything package.

Nathan Borggren, Hyoung-yoon Kim, Lihan Yao et al.

Monero is a popular crypto-currency which focuses on privacy. The blockchain uses cryptographic techniques to obscure transaction values as well as a `ring confidential transaction' which seeks to hide a real transaction among a variable number of spoofed transactions. We have developed training sets of simulated blockchains of 10 and 50 agents, for which we have control over the ground truth and keys, in order to test these claims. We featurize Monero transactions by characterizing the local structure of the public-facing blockchains and use labels obtained from the simulations to perform machine learning. Machine Learning of our features on the simulated blockchain shows that the technique can be used to aide in identifying individuals and groups, although it did not successfully reveal the hidden transaction values. We apply the technique on the real Monero blockchain to identify ShapeShift transactions, a cryptocurrency exchange that has leaked information through their API providing labels for themselves and their users.