Continuous monitoring of glacier calving fronts is essential for sea level rise projections. This study benchmarks Deep Learning systems for front delineation in Synthetic Aperture Radar imagery. While Deep Learning systems exhibit errors up to 221 m, human annotators deviate by only 38 m, underscoring the need for further research.

Aim: With the advent of LLMs, sophisticated agentic program repair has become viable at large organizations with large codebases. In this work, we develop an Engineering Agent that fixes the source code based on test failures at scale across diverse software offerings internally. Method: Using Llama as the base, we employ the ReAct harness to develop an agent. We start with a test failure that was triaged by a rule-based test failure bot. We then set up an agentic harness and allow the agent to reason and run a set of 15 actions from reading a file to generating a patch. We provide feedback to the agent through static analysis and test failures so it can refine its solution. We leverage an LLM-as-a-Judge to ensure that the patch conforms to the standards followed by a human review to land fixes. Benchmark Findings: We curated offline benchmarks for our patch generator, the Engineering Agent loop, and the LLM-as-a-Judge. In offline evaluations we found that a specialized 70B model is highly competitive with the much larger but vanilla Llama-405B. In an ablation study, we found that the ReAct harness (neural model) benefited from the symbolic information from static analysis tools and test execution traces. A model that strikes a balance between the solve rate and error rate vs the cost and latency has a benchmark solve rate of 42.3% using an average 11.8 feedback iterations. Production Findings: In a three month period, 80% of the generated fixes were reviewed, of which 31.5% were landed (25.5% of the total number of generated fixes). Feedback from Engineers: We used open coding to extract qualitative themes from engineers' feedback. We saw positive feedback in the form of quick approvals, gratitude, and surprise. We also found mixed feedback when the Engineering Agent's solution was partially correct and it served as a good starting point.

Social media platforms play an increasingly important role as forums for public discourse. Many platforms use recommendation algorithms that funnel users to online groups with the goal of maximizing user engagement, which many commentators have pointed to as a source of polarization and misinformation. Understanding the role of NLP in recommender systems is an interesting research area, given the role that social media has played in world events. However, there are few standardized resources which researchers can use to build models that predict engagement with online groups on social media; each research group constructs datasets from scratch without releasing their version for reuse. In this work, we present a dataset drawn from posts and comments on the online message board Reddit. We develop baseline models for recommending subreddits to users, given the user’s post and comment history. We also study the behavior of our recommender models on subreddits that were banned in June 2020 as part of Reddit’s efforts to stop the dissemination of hate speech.

New tools for raw data exploration and characterization of “big data” sets are required to suggest initial hypotheses for testing. The widespread use and adoption of web-based geo maps have provided a familiar set of interactions for exploring extremely large geo data spaces and can be applied to similarly large abstract data spaces. Building on these techniques, a tile based visual analytics system (TBVA) was developed that demonstrates interactive visualization for a one billion point Twitter dataset. TBVA enables John Tukey-inspired exploratory data analysis to be performed on massive data sets of effectively unlimited size.

One of the problems with mobile media devices is that they may distract users during critical everyday tasks, such as navigating the streets of a busy city. We addressed this issue in the design of eyeLook: a platform for attention sensitive mobile computing. eyeLook appliances use embedded low cost eyeCONTACT sensors (ECS) to detect when the user looks at the display. We discuss two eyeLook applications, seeTV and seeTXT, that facilitate courteous media consumption in mobile contexts by using the ECS to respond to user attention. seeTV is an attentive mobile video player that automatically pauses content when the user is not looking. seeTXT is an attentive speed reading application that flashes words on the display, advancing text only when the user is looking. By making mobile media devices sensitive to actual user attention, eyeLook allows applications to gracefully transition users between consuming media, and managing life.

No abstract available.

We present ECSGlasses: wearable eye contact sensing glasses that detect human eye contact. ECSGlasses report eye contact to digital devices, appliances and EyePliances in the user's attention space. Devices use this attentional cue to engage in a more sociable process of turn taking with users. This has the potential to reduce inappropriate intrusions, and limit their disruptiveness. We describe new prototype systems, including the Attentive Messaging Service (AMS), the Attentive Hit Counter, the first person attentive camcorder eyeBlog, and an updated Attentive Cell Phone. We also discuss the potential of these devices to open new windows of interaction using attention as a communication modality. Further, we present a novel signal-encoding scheme to uniquely identify EyePliances and users wearing ECSGlasses in multiparty scenarios.