Multimodal Large Language Models (MLLMs) have shown strong capabilities in image understanding, motivating recent efforts to extend them to video reasoning. However, existing Video LLMs struggle in online streaming scenarios, where long temporal context must be preserved under strict memory constraints. We propose WAT (Watching Before Thinking), a two-stage framework for online video reasoning. WAT separates processing into a query-independent watching stage and a query-triggered thinking stage. The watching stage builds a hierarchical memory system with a Short-Term Memory (STM) that buffers recent frames and a fixed-capacity Long-Term Memory (LTM) that maintains a diverse summary of historical content using a redundancy-aware eviction policy. In the thinking stage, a context-aware retrieval mechanism combines the query with the current STM context to retrieve relevant historical frames from the LTM for cross-temporal reasoning. To support training for online video tasks, we introduce WAT-85K, a dataset containing streaming-style annotations emphasizing real-time perception, backward tracing, and forecasting. Experiments show that WAT achieves state-of-the-art performance on onlin
What if a browser agent could learn your work simply by watching you do it? We present cotomi Act, a browser-based computer-using agent that combines reliable multi-step task execution with persistent organizational knowledge learned from user behavior. For execution, an agent scaffold with adaptive lazy observation, verbal-diff-based history compression, coarse-grained actions, and test-time scaling via best-of-N action selection achieves 80.4% on the 179-task WebArena human-evaluation subset, exceeding the reported 78.2% human baseline. For organizational knowledge, a behavior-to-knowledge pipeline passively observes the user's browsing and progressively abstracts it into artifacts (task boards, wiki) exposed through a shared workspace editable by both user and agent. A controlled proxy evaluation confirms that task success improves as behavior-derived knowledge accumulates. In our live demonstration, attendees interact with the system in a real browser, issuing tasks and observing end-to-end autonomous execution and shared knowledge management.
While videos have become increasingly prevalent in delivering information across different educational and professional contexts, individuals with ADHD often face attention challenges when watching informational videos due to the dynamic, multimodal, yet potentially distracting video elements. To understand and address this critical challenge, we designed FocusView, a video customization interface that allows viewers with ADHD to customize informational videos from different aspects. We evaluated FocusView with 12 participants with ADHD and found that FocusView significantly improved the viewability of videos by reducing distractions. Through the study, we uncovered participants' diverse perceptions of video distractions (e.g., background music as a distraction vs. stimulation boost) and their customization preferences, highlighting unique ADHD-relevant needs in designing video customization interfaces (e.g., reducing the number of options to avoid distraction caused by customization itself). We further derived design considerations for future video customization systems for the ADHD community.
Since the COVID-19 pandemic, online lectures have spread rapidly and many students are satisfied with them. However, one challenge remains the loss of concentration due to the lack of students' copresence. Our previous work suggests that presenting 3D characters with appropriate actions has the potential to improve concentration in online lectures. Nevertheless, an effective combination of actions has not yet been identified. In this study, we developed a lecture watching system that presents a 3D virtual classroom using a naked-eye 3D display. The system includes student characters that show copresence with various actions such as nodding, notetaking, and sleeping. An evaluation experiment was conducted with two conditions; (1) student characters perform only positive actions and (2) both positive and negative actions. The results, analyzed using posture and notetaking behavior as key indicators, suggest that the system can help to maintain concentration when the student characters perform both positive and negative actions, rather than only positive ones. These findings provide promising strategies for maintaining student focus in on-demand lectures and contribute to the developm
Watching TV not only provides news information but also gives an opportunity for different generations to communicate. With the proliferation of smartphones, PC, and the Internet, increase the opportunities for communication in front of the television is also likely to diminish. This has led to some problems further from face-to-face such as a lack of self-control and insufficient development of communication skills. This paper proposes a TV-watching companion robot with open-domain chat ability. The robot contains two modes: TV-watching mode and conversation mode. In TV-watching mode, the robot first extracts keywords from the TV program and then generates the disclosure utterances based on the extracted keywords as if enjoying the TV program. In the conversation mode, the robot generates question utterances with keywords in the same way and then employs a topics-based dialog management method consisting of multiple dialog engines for rich conversations related to the TV program. We conduct the initial experiments and the result shows that all participants from the three groups enjoyed talking with the robot, and the question about their interests in the robot was rated 6.5/7-leve
Prior work on dark patterns, or manipulative online interfaces, suggests they have potentially detrimental effects on user autonomy. Dark pattern features, like those designed for attention capture, can potentially extend platform sessions beyond that users would have otherwise intended. Existing research, however, has not formally measured the quantitative effects of these features on user engagement in subscription video-on-demand platforms (SVODs). In this work, we conducted an experimental study with 76 Netflix users in the US to analyze the impact of a specific attention capture feature, autoplay, on key viewing metrics. We found that disabling autoplay on Netflix significantly reduced key content consumption aggregates, including average daily watching and average session length, partly filling the evidentiary gap regarding the empirical effects of dark pattern interfaces. We paired the experimental analysis with users' perceptions of autoplay and their viewing behaviors, finding that participants were split on whether the effects of autoplay outweigh its benefits, albeit without knowledge of the study findings. Our findings strengthen the broader argument that manipulative i
The possibility of watching the Universe expand in real time and in a model-independent way, first envisaged by Allan Sandage more than 60 years ago and known as the redshift drift, is within reach of forthcoming astrophysical facilities, particularly the Extremely Large Telescope (ELT) and the Square Kilometre Array Observatory (SKAO). The latter, probing lower redshifts, enables us to watch the Universe's acceleration era in real time, while the former does the same for the matter era. We use Fisher Matrix Analysis techniques, which we show to give comparable results to those of a Markov Chain Monte Carlo approach, to discuss forecasts for SKAO measurements of the redshift drift and their cosmological impact. We consider specific fiducial cosmological models but mainly rely on a more agnostic cosmographic series (which includes the deceleration and jerk parameters), and we also discuss prospects for measurements of the drift of the drift. Overall, our analysis shows that SKAO measurements, with a reasonable amount of observing time, can provide a competitive probe of the low-redshift accelerating Universe.
Despite knowing the reality of three-dimensional (3D) technology in the form of eye fatigue, this technology continues to be retained by people (especially the young community). To check what happens before and after watching a 2D and 3D movie and how this condition influences the human brain's power spectrum density (PSD), a five-member test group was arranged. In this study, electroencephalogram (EEG) was used as a neuroimaging method. EEG recordings of five individuals were taken both before and after watching 2D and 3D movies. After 2D/3D EEG recording, this record was divided into three stages for analysis. These stages consisted of Relax, Watching, and Rest. This benchmarking analysis included I) before and after watching the 2D movie (R2b and R2a), II) before and after watching the 3D movie (R3b and R3a), and III) after watching the 2D/3D movie (R2a and R3a). In the Relax and Rest stages, the 2D/3D EEG power differences in all channels of brain regions for the five EEG bands, including delta (δ), theta (θ), alpha (α), beta (\b{eta}), and gamma (γ), were analyzed and compared.
Customer churn prediction is a valuable task in many industries. In telecommunications it presents great challenges, given the high dimensionality of the data, and how difficult it is to identify underlying frustration signatures, which may represent an important driver regarding future churn behaviour. Here, we propose a novel Bayesian hierarchical joint model that is able to characterise customer profiles based on how many events take place within different television watching journeys, and how long it takes between events. The model drastically reduces the dimensionality of the data from thousands of observations per customer to 11 customer-level parameter estimates and random effects. We test our methodology using data from 40 BT customers (20 active and 20 who eventually cancelled their subscription) whose TV watching behaviours were recorded from October to December 2019, totalling approximately half a million observations. Employing different machine learning techniques using the parameter estimates and random effects from the Bayesian hierarchical model as features yielded up to 92\% accuracy predicting churn, associated with 100\% true positive rates and false positive rat
We introduce the notion of watching systems in graphs, which is a generalization of that of identifying codes. We give some basic properties of watching systems, an upper bound on the minimum size of a watching system, and results on the graphs which achieve this bound; we also study the cases of the paths and cycles, and give complexity results.
We demonstrate in this letter a unique approach for watching outside while hiding in a carpet cloaking based on transformation optics. Unlike conventional carpet cloaking, which screens all the incident electromagnetic waves, we break the cloak and allow incident light get into the carpet. Hence outside information is detected inside the cloak. To recover the invisible cloaking, complementary techniques are applied in the broken space. Consequently, a hiding-inside-and-watching-outside (HIWO) carpet cloak is sewed, which works as a perfectly invisible cloaking and allows surveillance of the outside at the same time. Our work provides a strategy for ideal cloak with "hiding" and "watching" functions simultaneously.
Deep visuomotor policy learning, which aims to map raw visual observation to action, achieves promising results in control tasks such as robotic manipulation and autonomous driving. However, it requires a huge number of online interactions with the training environment, which limits its real-world application. Compared to the popular unsupervised feature learning for visual recognition, feature pretraining for visuomotor control tasks is much less explored. In this work, we aim to pretrain policy representations for driving tasks by watching hours-long uncurated YouTube videos. Specifically, we train an inverse dynamic model with a small amount of labeled data and use it to predict action labels for all the YouTube video frames. A new contrastive policy pretraining method is then developed to learn action-conditioned features from the video frames with pseudo action labels. Experiments show that the resulting action-conditioned features obtain substantial improvements for the downstream reinforcement learning and imitation learning tasks, outperforming the weights pretrained from previous unsupervised learning methods and ImageNet pretrained weight. Code, model weights, and data ar
Video understanding is being rapidly transformed by multimodal large language models (MLLMs), as research moves from short clips to long, multimodal, and knowledge-intensive video scenarios. These scenarios require models to handle sparse evidence, long-range dependencies, multimodal alignment, and reliable inference under limited computational budgets. This work presents a human-view perspective on LLM-based video understanding, organized around three functional abilities: watching, remembering, and reasoning. Rather than treating video tasks as isolated benchmarks, this view provides a unified structure for analyzing how video MLLMs acquire evidence, preserve context, and produce grounded outputs. We introduce a formulation that characterizes video understanding systems by their perceptual representations, memory states, reasoning traces, and final predictions. Based on this formulation, we identify challenges in spatio-temporal perception, efficient long-video processing, memory modeling, streaming understanding, and faithful reasoning. Representative methods are organized by their roles in video MLLM systems. Watching covers fine-grained, comprehensive, audio-visual, and effici
The significance of estimating video watch time has been highlighted by the rising importance of (short) video recommendation, which has become a core product of mainstream social media platforms. Modeling video watch time, however, has been challenged by the complexity of user-video interaction, such as different user behavior modes in watching the recommended videos and varying watching probability over the video progress bar. Despite the importance and challenges, existing literature on modeling video watch time mostly focuses on relatively black-box mechanical enhancement of the classical regression/classification losses, without factoring in user behavior in a principled manner. In this paper, we for the first time take on a user-centric perspective to model video watch time, from which we propose a white-box statistical framework that directly translates various user behavior assumptions in watching (short) videos into statistical watch time models. These behavior assumptions are portrayed by our domain knowledge on users' behavior modes in video watching. We further employ bucketization to cope with user's non-stationary watching probability over the video progress bar, whic
We present a design rationale, embedding model, and interactive visual-analysis system for exploring large wristwatch collections through heterogeneous visual and semantic attributes. The system addresses a common limitation of catalog and e-commerce interfaces: users can filter by metadata, but they receive little support for open-ended exploration of visual similarity, stylistic alternatives, and mixed aesthetic-functional criteria. We therefore represent watches with separate attribute graphs for dial color and dial design, while using watch type as an explicit semantic organizer. Dials are segmented with a U-Net, watch types are predicted with a Vision Transformer, colors are represented through a shared CIELAB reference palette, and dial structure is described with a gradient-based image descriptor. We extend UMAP by combining attribute-specific neighborhood graphs in a unified probabilistic objective and by adding a class-aware layout term that separates global type structure from local visual neighborhoods. The resulting map is exposed in an interactive interface with spatial navigation, metadata filtering, detail inspection, and search-by-example insertion. We evaluate the
In the video recommendation, watch time is commonly adopted as an indicator of user interest. However, watch time is not only influenced by the matching of users' interests but also by other factors, such as duration bias and noisy watching. Duration bias refers to the tendency for users to spend more time on videos with longer durations, regardless of their actual interest level. Noisy watching, on the other hand, describes users taking time to determine whether they like a video or not, which can result in users spending time watching videos they do not like. Consequently, the existence of duration bias and noisy watching make watch time an inadequate label for indicating user interest. Furthermore, current methods primarily address duration bias and ignore the impact of noisy watching, which may limit their effectiveness in uncovering user interest from watch time. In this study, we first analyze the generation mechanism of users' watch time from a unified causal viewpoint. Specifically, we considered the watch time as a mixture of the user's actual interest level, the duration-biased watch time, and the noisy watch time. To mitigate both the duration bias and noisy watching, we
The two-watched literal scheme, a core component of efficient CDCL (Conflict-Driven Clause Learning) implementations for propositional logic, is extended to first-order logic. Given a set of first-order clauses and a set of ground literals, our lifted two-watched literal scheme efficiently detects all propagating and false clauses with respect to the ground literals. We present the algorithm as a system of rules and prove its soundness and completeness. Additionally, we provide an implementation of the two-watched literal scheme, which outperforms a standard dynamic programming approach for detecting propagatable literals and conflicts, especially when dealing with long clauses.
Background: With the popularity of live streaming platforms at an all-time high, and many people turning to alternative venues for educational needs, this full research paper explores the viewership habits of software and game development live streams through the lens of informal education opportunities. Purpose: We investigate why developers watch software and game development live streams to understand the educational and social benefits they derive from this emerging form of informal learning. Methods: We implement a mixed-methods study combining survey data from 39 viewers and nine semi-structured interviews to analyze motivations, perceptions, and outcomes of watching development live streams. Findings: This research finds that viewers are motivated by both educational and social factors, with community engagement and informal mentorship as key motivations. Additionally, we find that technical learning draws initial interest, but social connections and co-working aspects sustain long-term engagement. Implications: Live streaming serves as a valuable informal learning tool that combines self-directed technical education with community support, which suggests that developers can
In online video platforms, accurate watch time prediction has become a fundamental and challenging problem in video recommendation. Previous research has revealed that the accuracy of watch time prediction highly depends on both the transformation of watch-time labels and the decomposition of the estimation process. TPM (Tree based Progressive Regression Model) achieves State-of-the-Art performance with a carefully designed and effective decomposition paradigm. TPM discretizes the watch time into several ordinal intervals and organizes them into a binary decision tree, where each node corresponds to a specific interval. At each non-leaf node, a binary classifier is used to determine the specific interval in which the watch time variable most likely falls, based on the prediction outcome at its parent node. The tree structure is central to TPM, as it defines the decomposition of watch time estimation and how ordinal intervals are discretized. However, TPM uses a predefined full binary tree, which may be sub-optimal for two reasons. First, full binary trees imply equal partitioning of the watch time space, which may fail to capture the complexity of real-world distributions. Second,
In video recommendation, a critical component that determines the system's recommendation accuracy is the watch-time prediction module, since how long a user watches a video directly reflects personalized preferences. One of the key challenges of this problem is the user's stochastic watch-time behavior. To improve the prediction accuracy for such an uncertain behavior, existing approaches show that one can either reduce the noise through duration bias modeling or formulate a distribution modeling task to capture the uncertainty. However, the uncontrolled uncertainty is not always equally distributed across users and videos, inducing a balancing paradox between the model accuracy and the ability to capture out-of-distribution samples. In practice, we find that the uncertainty of the watch-time prediction model also provides key information about user behavior, which, in turn, could benefit the prediction task itself. Following this notion, we derive an explicit uncertainty modeling strategy for the prediction model and propose an adversarial optimization framework that can better exploit the user watch-time behavior. This framework has been deployed online on an industrial video sh