Riku Arakawa

Aarohi Srivastava, Abhinav Rastogi, Abhishek Rao et al. · allen-ai, amazon-science

Language models demonstrate both quantitative improvement and new qualitative capabilities with increasing scale. Despite their potentially transformative impact, these new capabilities are as yet poorly characterized. In order to inform future research, prepare for disruptive new model capabilities, and ameliorate socially harmful effects, it is vital that we understand the present and near-future capabilities and limitations of language models. To address this challenge, we introduce the Beyond the Imitation Game benchmark (BIG-bench). BIG-bench currently consists of 204 tasks, contributed by 450 authors across 132 institutions. Task topics are diverse, drawing problems from linguistics, childhood development, math, common-sense reasoning, biology, physics, social bias, software development, and beyond. BIG-bench focuses on tasks that are believed to be beyond the capabilities of current language models. We evaluate the behavior of OpenAI's GPT models, Google-internal dense transformer architectures, and Switch-style sparse transformers on BIG-bench, across model sizes spanning millions to hundreds of billions of parameters. In addition, a team of human expert raters performed all tasks in order to provide a strong baseline. Findings include: model performance and calibration both improve with scale, but are poor in absolute terms (and when compared with rater performance); performance is remarkably similar across model classes, though with benefits from sparsity; tasks that improve gradually and predictably commonly involve a large knowledge or memorization component, whereas tasks that exhibit "breakthrough" behavior at a critical scale often involve multiple steps or components, or brittle metrics; social bias typically increases with scale in settings with ambiguous context, but this can be improved with prompting.

Vimal Mollyn, Riku Arakawa, Mayank Goel et al.

Tracking body pose on-the-go could have powerful uses in fitness, mobile gaming, context-aware virtual assistants, and rehabilitation. However, users are unlikely to buy and wear special suits or sensor arrays to achieve this end. Instead, in this work, we explore the feasibility of estimating body pose using IMUs already in devices that many users own -- namely smartphones, smartwatches, and earbuds. This approach has several challenges, including noisy data from low-cost commodity IMUs, and the fact that the number of instrumentation points on a users body is both sparse and in flux. Our pipeline receives whatever subset of IMU data is available, potentially from just a single device, and produces a best-guess pose. To evaluate our model, we created the IMUPoser Dataset, collected from 10 participants wearing or holding off-the-shelf consumer devices and across a variety of activity contexts. We provide a comprehensive evaluation of our system, benchmarking it on both our own and existing IMU datasets.

Riku Arakawa, Hiromu Yakura, Masataka Goto

CatAlyst uses generative models to help workers' progress by influencing their task engagement instead of directly contributing to their task outputs. It prompts distracted workers to resume their tasks by generating a continuation of their work and presenting it as an intervention that is more context-aware than conventional (predetermined) feedback. The prompt can function by drawing their interest and lowering the hurdle for resumption even when the generated continuation is insufficient to substitute their work, while recent human-AI collaboration research aiming at work substitution depends on a stable high accuracy. This frees CatAlyst from domain-specific model-tuning and makes it applicable to various tasks. Our studies involving writing and slide-editing tasks demonstrated CatAlyst's effectiveness in helping workers swiftly resume tasks with a lowered cognitive load. The results suggest a new form of human-AI collaboration where large generative models publicly available but imperfect for each individual domain can contribute to workers' digital well-being.

Taejun Kim, Vimal Mollyn, Riku Arakawa et al.

We present a new and accurate approach for gaze estimation on consumer computing devices. We take advantage of continued strides in the quality of user-facing cameras found in e.g., smartphones, laptops, and desktops - 4K or greater in high-end devices - such that it is now possible to capture the 2D reflection of a device's screen in the user's eyes. This alone is insufficient for accurate gaze tracking due to the near-infinite variety of screen content. Crucially, however, the device knows what is being displayed on its own screen - in this work, we show this information allows for robust segmentation of the reflection, the location and size of which encodes the user's screen-relative gaze target. We explore several strategies to leverage this useful signal, quantifying performance in a user study. Our best performing model reduces mean tracking error by ~8% compared to a baseline appearance-based model. A supplemental study reveals an additional 10-20% improvement if the gaze-tracking camera is located at the bottom of the device.

Riku Arakawa, Hiromu Yakura, Mayank Goel

We routinely perform procedures (such as cooking) that include a set of atomic steps. Often, inadvertent omission or misordering of a single step can lead to serious consequences, especially for those experiencing cognitive challenges such as dementia. This paper introduces PrISM-Observer, a smartwatch-based, context-aware, real-time intervention system designed to support daily tasks by preventing errors. Unlike traditional systems that require users to seek out information, the agent observes user actions and intervenes proactively. This capability is enabled by the agent's ability to continuously update its belief in the user's behavior in real-time through multimodal sensing and forecast optimal intervention moments and methods. We first validated the steps-tracking performance of our framework through evaluations across three datasets with different complexities. Then, we implemented a real-time agent system using a smartwatch and conducted a user study in a cooking task scenario. The system generated helpful interventions, and we gained positive feedback from the participants. The general applicability of PrISM-Observer to daily tasks promises broad applications, for instance, including support for users requiring more involved interventions, such as people with dementia or post-surgical patients.

Riku Arakawa, Hiromu Yakura

Recent organizations have started to adopt AI-based decision support tools to optimize human resource development practices, while facing various challenges of using AIs in highly contextual and sensitive domains. We present our case study that aims to help professional assessors make decisions in human assessment, in which they conduct interviews with assessees and evaluate their suitability for certain job roles. Our workshop with two industrial assessors elucidated troubles they face (i.e., maintaining stable and non-subjective observation of assessees' behaviors) and derived requirements of AI systems (i.e., extracting their nonverbal cues from interview videos in an interpretable manner). In response, we employed an unsupervised anomaly detection algorithm using multimodal behavioral features such as facial keypoints, body and head pose, and gaze. The algorithm extracts outlier scenes from the video based on behavioral features as well as informing which feature contributes to the outlierness. We first evaluated how the assessors would perceive the extracted cues and discovered that the algorithm is useful in suggesting scenes to which assessors would pay attention, thanks to its interpretability. Then, we developed an interface prototype incorporating the algorithm and had six assessors use it for their actual assessment. Their comments revealed the effectiveness of introducing unsupervised anomaly detection to enhance their feeling of confidence and objectivity of the assessment along with potential use scenarios of such AI-based systems in human assessment. Our approach, which builds on top of the idea of separating observation and interpretation in human-AI collaboration, will facilitate human decision making in highly contextual domains, such as human assessment, while keeping their trust in the system.

Riku Arakawa, Hiromu Yakura

In this paper, we discuss the potential of applying unsupervised anomaly detection in constructing AI-based interactive systems that deal with highly contextual situations, i.e., human-human communication, in collaboration with domain experts. We reached this approach of utilizing unsupervised anomaly detection through our experience of developing a computational support tool for executive coaching, which taught us the importance of providing interpretable results so that expert coaches can take both the results and contexts into account. The key idea behind this approach is to leave room for expert coaches to unleash their open-ended interpretations, rather than simplifying the nature of social interactions to well-defined problems that are tractable by conventional supervised algorithms. In addition, we found that this approach can be extended to nurturing novice coaches; by prompting them to interpret the results from the system, it can provide the coaches with educational opportunities. Although the applicability of this approach should be validated in other domains, we believe that the idea of leveraging unsupervised anomaly detection to construct AI-based interactive systems would shed light on another direction of human-AI communication.

Prasoon Patidar, Riku Arakawa, Ricardo Graça et al.

Deploying human activity recognition (HAR) at home is still rare because sensor signals vary wildly across houses, people, and time, essentially requiring in-situ data collection and training. Prior approaches use cameras to generate training labels for privacy-preserving sensors (LiDAR, RADAR, Thermal), but this forces sensors to detect predefined activities that cameras can see yet the sensors themselves cannot reliably distinguish. In this work, we introduce OrganicHAR, an activity discovery framework that inverts this relationship by placing sensor capabilities at the center of activity discovery. Our approach identifies naturally occurring signal patterns using privacy-preserving sensors, leverages Vision Language Models (VLMs) only during these key moments for scene understanding, and discovers discrete activity labels at granularities that these sensors can reliably detect. Our evaluation with 12 participants demonstrates OrganicHAR's effectiveness: it achieves 79% accuracy for coarse (4-5) activities using only basic ambient sensors (radar, lidar, thermal arrays), and 73% accuracy for fine-grained (8-9) activities when a wearable IMU, depth, and pose sensor are added. OrganicHAR maintains 77% accuracy on average across configurations while discovering 4-8 categories per user (15 across all users) tailored to each environment and sensor capabilities. By triggering video processing only at key moments identified by local sensors, we reduce queries to VLM by 90%, enabling practical and privacy-preserving activity recognition in natural settings.

Nicholas Magal, Minh Tran, Riku Arakawa et al.

This paper aims to document an effective way to improve multimodal co-learning by using aggressive modality dropout. We find that by using aggressive modality dropout we are able to reverse negative co-learning (NCL) to positive co-learning (PCL). Aggressive modality dropout can be used to "prep" a multimodal model for unimodal deployment, and dramatically increases model performance during negative co-learning, where during some experiments we saw a 20% gain in accuracy. We also benchmark our modality dropout technique against PCL to show that our modality drop out technique improves co-learning during PCL, although it does not have as much as an substantial effect as it does during NCL. Github: https://github.com/nmagal/modality_drop_for_colearning

Riku Arakawa, Hiromu Yakura

Chatbots' role in fostering self-reflection is now widely recognized, especially in inducing users' behavior change. While the benefits of 24/7 availability, scalability, and consistent responses have been demonstrated in contexts such as healthcare and tutoring to help one form a new habit, their utilization in coaching necessitating deeper introspective dialogue to induce leadership growth remains unexplored. This paper explores the potential of such a chatbot powered by recent Large Language Models (LLMs) in collaboration with professional coaches in the field of executive coaching. Through a design workshop with them and two weeks of user study involving ten coach-client pairs, we explored the feasibility and nuances of integrating chatbots to complement human coaches. Our findings highlight the benefits of chatbots' ubiquity and reasoning capabilities enabled by LLMs while identifying their limitations and design necessities for effective collaboration between human coaches and chatbots. By doing so, this work contributes to the foundation for augmenting one's self-reflective process with prevalent conversational agents through the human-in-the-loop approach.

Haozhe Zhou, Riku Arakawa, Yuvraj Agarwal et al.

IMUs are regularly used to sense human motion, recognize activities, and estimate full-body pose. Users are typically required to place sensors in predefined locations that are often dictated by common wearable form factors and the machine learning model's training process. Consequently, despite the increasing number of everyday devices equipped with IMUs, the limited adaptability has seriously constrained the user experience to only using a few well-explored device placements (e.g., wrist and ears). In this paper, we rethink IMU-based motion sensing by acknowledging that signals can be captured from any point on the human body. We introduce IMU over Continuous Coordinates (IMUCoCo), a novel framework that maps signals from a variable number of IMUs placed on the body surface into a unified feature space based on their spatial coordinates. These features can be plugged into downstream models for pose estimation and activity recognition. Our evaluations demonstrate that IMUCoCo supports accurate pose estimation in a wide range of typical and atypical sensor placements. Overall, IMUCoCo supports significantly more flexible use of IMUs for motion sensing than the state-of-the-art, allowing users to place their sensors-laden devices according to their needs and preferences. The framework also supports the ability to change device locations depending on the context and suggests placement depending on the use case.

Riku Arakawa, Kiyosu Maeda, Hiromu Yakura

Multimodal scene search of conversations is essential for unlocking valuable insights into social dynamics and enhancing our communication. While experts in conversational analysis have their own knowledge and skills to find key scenes, a lack of comprehensive, user-friendly tools that streamline the processing of diverse multimodal queries impedes efficiency and objectivity. To solve it, we developed Providence, a visual-programming-based tool based on design considerations derived from a formative study with experts. It enables experts to combine various machine learning algorithms to capture human behavioral cues without writing code. Our study showed its preferable usability and satisfactory output with less cognitive load imposed in accomplishing scene search tasks of conversations, verifying the importance of its customizability and transparency. Furthermore, through the in-the-wild trial, we confirmed the objectivity and reusability of the tool transform experts' workflow, suggesting the advantage of expert-AI teaming in a highly human-contextual domain.

Katsushi Ikeuchi, Naoki Wake, Riku Arakawa et al.

The paradigm of learning-from-observation (LfO) enables a robot to learn how to perform actions by observing human-demonstrated actions. Previous research in LfO have mainly focused on the industrial domain which only consist of the observable physical constraints between a manipulating tool and the robot's working environment. In order to extend this paradigm to the household domain which consists non-observable constraints derived from a human's common sense; we introduce the idea of semantic constraints. The semantic constraints are represented similar to the physical constraints by defining a contact with an imaginary semantic environment. We thoroughly investigate the necessary and sufficient set of contact state and state transitions to understand the different types of physical and semantic constraints. We then apply our constraint representation to analyze various actions in top hit household YouTube videos and real home cooking recordings. We further categorize the frequently appearing constraint patterns into physical, semantic, and multistage task groups and verify that these groups are not only necessary but a sufficient set for covering standard household actions. Finally, we conduct a preliminary experiment using textual input to explore the possibilities of combining verbal and visual input for recognizing the task groups. Our results provide promising directions for incorporating common sense in the literature of robot teaching.

Riku Arakawa, Hiromu Yakura

A growing number of people are using catch-up TV services rather than watching simultaneously with other audience members at the time of broadcast. However, computational support for such catching-up users has not been well explored. In particular, we are observing an emerging phenomenon in online media consumption experiences in which speculation plays a vital role. As the phenomenon of speculation implicitly assumes simultaneity in media consumption, there is a gap for catching-up users, who cannot directly appreciate the consumption experiences. This conversely suggests that there is potential for computational support to enhance the consumption experiences of catching-up users. Accordingly, we conducted a series of studies to pave the way for developing computational support for catching-up users. First, we conducted semi-structured interviews to understand how people are engaging with speculation during media consumption. As a result, we discovered the distinctive aspects of speculation-based consumption experiences in contrast to social viewing experiences sharing immediate reactions that have been discussed in previous studies. We then designed two prototypes for supporting catching-up users based on our quantitative analysis of Twitter data in regard to reaction- and speculation-based media consumption. Lastly, we evaluated the prototypes in a user experiment and, based on its results, discussed ways to empower catching-up users with computational supports in response to recent transformations in media consumption.

Riku Arakawa, Hiromu Yakura

Explicitly alerting users is not always an optimal intervention, especially when they are not motivated to obey. For example, in video-based learning, learners who are distracted from the video would not follow an alert asking them to pay attention. Inspired by the concept of Mindless Computing, we propose a novel intervention approach, Mindless Attractor, that leverages the nature of human speech communication to help learners refocus their attention without relying on their motivation. Specifically, it perturbs the voice in the video to direct their attention without consuming their conscious awareness. Our experiments not only confirmed the validity of the proposed approach but also emphasized its advantages in combination with a machine learning-based sensing module. Namely, it would not frustrate users even though the intervention is activated by false-positive detection of their attentive state. Our intervention approach can be a reliable way to induce behavioral change in human-AI symbiosis.

Naoki Wake, Riku Arakawa, Iori Yanokura et al.

A household robot is expected to perform various manipulative operations with an understanding of the purpose of the task. To this end, a desirable robotic application should provide an on-site robot teaching framework for non-experts. Here we propose a Learning-from-Observation (LfO) framework for grasp-manipulation-release class household operations (GMR-operations). The framework maps human demonstrations to predefined task models through one-shot teaching. Each task model contains both high-level knowledge regarding the geometric constraints and low-level knowledge related to human postures. The key idea is to design a task model that 1) covers various GMR-operations and 2) includes human postures to achieve tasks. We verify the applicability of our framework by testing an operational LfO system with a real robot. In addition, we quantify the coverage of the task model by analyzing online videos of household operations. In the context of one-shot robot teaching, the contribution of this study is a framework that 1) covers various GMR-operations and 2) mimics human postures during the operations.

Riku Arakawa, Shintaro Shiba

We demonstrate the first reinforcement-learning application for robots equipped with an event camera. Because of the considerably lower latency of the event camera, it is possible to achieve much faster control of robots compared with the existing vision-based reinforcement-learning applications using standard cameras. To handle a stream of events for reinforcement learning, we introduced an image-like feature and demonstrated the feasibility of training an agent in a simulator for two tasks: fast collision avoidance and obstacle tracking. Finally, we set up a robot with an event camera in the real world and then transferred the agent trained in the simulator, resulting in successful fast avoidance of randomly thrown objects. Incorporating event camera into reinforcement learning opens new possibilities for various robotics applications that require swift control, such as autonomous vehicles and drones, through end-to-end learning approaches.

Riku Arakawa, Hiromu Yakura

Executive coaching has been drawing more and more attention for developing corporate managers. While conversing with managers, coach practitioners are also required to understand internal states of coachees through objective observations. In this paper, we present REsCUE, an automated system to aid coach practitioners in detecting unconscious behaviors of their clients. Using an unsupervised anomaly detection algorithm applied to multimodal behavior data such as the subject's posture and gaze, REsCUE notifies behavioral cues for coaches via intuitive and interpretive feedback in real-time. Our evaluation with actual coaching scenes confirms that REsCUE provides the informative cues to understand internal states of coachees. Since REsCUE is based on the unsupervised method and does not assume any prior knowledge, further applications beside executive coaching are conceivable using our framework.

Riku Arakawa, Sosuke Kobayashi, Yuya Unno et al.

Exploration has been one of the greatest challenges in reinforcement learning (RL), which is a large obstacle in the application of RL to robotics. Even with state-of-the-art RL algorithms, building a well-learned agent often requires too many trials, mainly due to the difficulty of matching its actions with rewards in the distant future. A remedy for this is to train an agent with real-time feedback from a human observer who immediately gives rewards for some actions. This study tackles a series of challenges for introducing such a human-in-the-loop RL scheme. The first contribution of this work is our experiments with a precisely modeled human observer: binary, delay, stochasticity, unsustainability, and natural reaction. We also propose an RL method called DQN-TAMER, which efficiently uses both human feedback and distant rewards. We find that DQN-TAMER agents outperform their baselines in Maze and Taxi simulated environments. Furthermore, we demonstrate a real-world human-in-the-loop RL application where a camera automatically recognizes a user's facial expressions as feedback to the agent while the agent explores a maze.