Log anomaly detection (LogAD) is crucial for identifying failures and threats in large-scale computing and cyber-infrastructure systems. However, most existing LogAD approaches suffer from key limitations: they depend on slow and error-prone log parsing, employ tightly coupled end-to-end pipelines, often require supervision for improved detection performance, and rely on flawed single-pass evaluation protocols that fail to reflect the temporal dynamics of real-world online detection. These issues significantly hinder scalability, adaptability, and the practical deployment of solutions. To address these limitations, we introduce$\mathbf{K}^{\mathbf{4}}$(Knowing the Unknown by Knowing only the Known), a fully unsupervised, parser-independent, and representation-agnostic LogAD framework designed for high-performance online detection. At its core,$K^{4}$is grounded in a novel formulation based on representation-level typicality estimation, which transforms arbitrary log embeddings into compact and interpretable four-dimensional descriptors: Precision, Recall, Density, and Coverage (PRDC), which are swiftly computed via GPU-acceleration into geometric$k$-nearest neighbor statistics. These descriptors inform lightweight, modular detectors, including KDE, GMM, OCSVM, and a new adaptation of DeepSVDD, which enables efficient and accurate anomaly scoring without relying on structured formats or log representation retraining. To support realistic deployment scenarios, we also propose a principled streaming-faithful evaluation protocol that partitions datasets into fixed-size chunks and applies sliding-window sampling with strides to emulate online log ingestion, alleviating the performance overestimation and dataset undercoverage issues of prior single-pass evaluations and enabling reproducible benchmarking across datasets with varying anomaly densities. Using this setup, we conduct over$\mathbf{1 2 5, 0 0 0}$experiments across three real-world datasets (HDFS, BGL, Thunderbird), six pretrained embedding models, four detectors, and multiple training and log sampling configurations. Compared to six representative baseline methods spanning supervised, semi-supervised, self-supervised, and unsupervised paradigms,$K^{4}$consistently sets new state-of-the-art results (AUROC: 0.995-0.999, F1: 0.989-0.992) and outperforms all baselines by large margins, while keeping detector training under 4 seconds and per-sample inference latency as low as$4 \mu ~\mathrm{s}$, which are orders of magnitude faster than the most competitive alternatives.

Training Large Language Models(LLMs) is one of the most compute-intensive tasks in high-performance computing. Predicting end-to-end training time for multi-billion parameter models distributed across hundreds of GPUs remains challenging due to complex interactions between transformer components, parallelism strategies(data, model, pipeline, tensor), and multi-tier communication. Learned models require costly sampling, while analytical models often struggle with real-world network and hardware complexities. We address this by decomposing LLMs into core computational primitives and modeling them with: (1) operator-level decomposition for fine-grained analysis; (2) lightweight sampling based hardware-aware prediction models for key operations; (3) an end-to-end prediction system integrating these components across complex parallelization strategies. Crucially, our methodology has been validated on two large-scale HPC systems. Our framework achieves low average prediction errors-4.98% on Perlmutter(A100) and 9.38% on Vista(GH200)-for models up to 20B parameters across 128 GPUs. Importantly, it runs entirely on CPUs, enabling rapid iteration over hardware configurations and training strategies without costly on-cluster experimentation.

The Bitcoin Lightning Network is a Layer 2 payment protocol that addresses Bitcoin’s scalability by facilitating quick and cost-effective transactions through payment channels. This research explores the use of machine learning models to interpolate channel balances within the network, which can be used for optimizing the network’s pathfinding algorithms. While there has been much exploration in balance probing and multipath payment protocols, predicting channel balances using solely node and channel features remains an uncharted area. This paper evaluates the performance of several machine learning models against two heuristic baselines and investigates the predictive capabilities of various features. Our model performs favorably in experimental evaluation, reducing the relative error by 27% compared to an equal split baseline where both edges are assigned half of the channel capacity.

Large language models (LLMs) show remarkable promise for democratizing automated reasoning by generating formal specifications. However, a fundamental tension exists: LLMs are probabilistic, while formal verification demands deterministic guarantees. This paper addresses this epistemological gap by comprehensively investigating failure modes and uncertainty quantification (UQ) in LLM-generated formal artifacts. Our systematic evaluation of five frontier LLMs reveals Satisfiability Modulo Theories (SMT) based autoformalization's domain-specific impact on accuracy (from +34.8\% on logical tasks to -44.5\% on factual ones), with known UQ techniques like the entropy of token probabilities failing to identify these errors. We introduce a probabilistic context-free grammar (PCFG) framework to model LLM outputs, yielding a refined uncertainty taxonomy. We find uncertainty signals are task-dependent (e.g., grammar entropy for logic, AUROC>0.93). Finally, a lightweight fusion of these signals enables selective verification, drastically reducing errors (14-100\%) with minimal abstention, transforming LLM-driven formalization into a reliable engineering discipline.

A new time switching protocol referred to as TS- APS is proposed for the energy harvesting (EH) relay networks based on the combination of conventional time switching (TS) protocol and adaptive power splitting (APS). In the framework of the decode and forward (DF) relaying with multiple relays, we derive the closed-form expression for the outage probability of the proposed scheme and discuss its effective transmission rate considering the random relay selection (RRS) and opportunistic relay selection (ORS) scenarios. Through numerical analysis, we show that the proposed scheme can achieve better outage performance and the effective transmission rate as compared to the conventional TS and power splitting (PS) schemes proposed in the literature, irrespective of relay selection scenarios. By properly selecting time switching (TS) ratio, we demonstrate that the effective transmission rate of the proposed protocol can be optimized only with partial knowledge of channel gains in the network, which makes the proposed approach suitable for the distributed relay selection scenarios.

Video usage in educational environments has increased significantly with the expansion of online learning opportunities. However, current video-based applications used for learning are designed for passive consumption of video content and do not support small-group collaboration. Specifically, they lack interaction affordances for active and collaborative learning through video materials. In this paper, we present the affordances and user experience of a web-based video application called Video Collaboratory that is designed with multiple affordances for group-based learning around video content. We surveyed students and analyzed the use of the Video Collaboratory in an HCI course. Students found the Video Collaboratory helpful because they could learn from their peers. Students also used the application beyond the activities required for credit.

The dance production process is strongly influenced within the physical rehearsal space by social context factors and dynamics, such as intimacy of bodies, gender distribution, and the hierarchy of choreographers and dancers. Introducing online tools for asynchronous collaboration can change the traditional dance production process and impact the social dynamics of the group. We developed and deployed the Choreographer's Notebook, a web-based, collaborative, multi-modal annotation tool used in the creative process of making dance. We collected usage logs and choreographer reflections on the use of this tool, along with conducting interviews and focus groups, from the interdisciplinary perspectives of both technologists and choreographers involved in the project. We describe the socio-technical impacts of the Choreographer's Notebook based on the results of its usage in three dance productions. We analyze these case studies through various contextual lenses and provide a visualization of how the choreographic correction process evolved.

We present a collaborative tool for choreographers and dancers, the Choreographer's Notebook, which allows multimodal annotation of rehearsal videos during the dance production process. The dance production process is a creative process in which exploration and expression are paramount and we describe issues we have observed over four different dance production cycles. The Choreographer's Notebook addresses the recurring problem of scarcity of rehearsal time and space, which is a limiting factor for exploration and expression. The tool is a web-based video application that enables choreographers to provide multimodal, context-sensitive instructions and feedback to the dancers outside of formal rehearsal time. The novelty of this tool is its ability to provide a multimodal collaborative video-based workflow. This article also presents a thorough ethnographic investigation of the dance production process and results from the use of the Choreographer's Notebook as an artifact within that process.