Learning a Hierarchical Intent Model for Next-Item Recommendation

A session-based recommender system (SBRS) captures users’ evolving behaviors and recommends the next item by profiling users in terms of items in a session. User intent and user preference are two factors affecting his (her) decisions. Specifically, the former narrows the selection scope to some item types, while the latter helps to compare items of the same type. Most SBRSs assume one arbitrary user intent dominates a session when making a recommendation. However, this oversimplifies the reality that a session may involve multiple types of items conforming to different intents. In current SBRSs, items conforming to different user intents have cross-interference in profiling users for whom only one user intent is considered. Explicitly identifying and differentiating items conforming to various user intents can address this issue and model rich contextual information of a session. To this end, we design a framework modeling user intent and preference explicitly, which empowers the two factors to play their distinctive roles. Accordingly, we propose a key-array memory network (KA-MemNN) with a hierarchical intent tree to model coarse-to-fine user intents. The two-layer weighting unit (TLWU) in KA-MemNN detects user intents and generates intent-specific user profiles. Furthermore, the hierarchical semantic component (HSC) integrates multiple sets of intent-specific user profiles along with different user intent distributions to model a multi-intent user profile. The experimental results on real-world datasets demonstrate the superiority of KA-MemNN over selected state-of-the-art methods.

Feasibility of Using Wearable EMG Armbands combined with Unsupervised Transfer Learning for Seamless Myoelectric Control

Myoelectric control uses electromyography (EMG) signals as human-originated input to enable intuitive interfaces with machines. As such, recent rehabilitation robotics employs myoelectric control to autonomously classify user intent or operation mode using machine learning. However, performance in such applications inherently suffers from the non-stationarity of EMG signals across measurement conditions. Current laboratory-based solutions rely on careful, time-consuming control of the recordings or periodic recalibration, impeding real-world deployment. We propose that robust yet seamless myoelectric control can be achieved using a low-end, easy-to-don and doff wearable EMG sensor combined with unsupervised transfer learning. Here, we test the feasibility of one such application using a consumer-grade sensor (Myo armband, 8 EMG channels @ 200 Hz) for gesture classification across measurement conditions using an existing dataset: 5 users x 10 days x 3 sensor locations. Specifically, we first train a deep neural network using Temporal-Spatial Descriptors (TSD) with labeled source data from any particular user, day, or location. We then apply the Self-Calibrating Asynchronous Domain Adversarial Neural Network (SCADANN), which automatically adjusts the trained TSD to improve classification performance for unlabeled target data from a different user, day, or sensor location. Compared to the original TSD, SCADANN improves accuracy by 12±5.2% (avg±sd), 9.6±5.0%, and 8.6±3.3% across all possible user-to-user, day-to-day, and location-to-location cases, respectively. In one best-case scenario, accuracy improves by 26% (from 67% to 93%), whereas sometimes the gain is modest (e.g., from 76% to 78%). We also show that the performance of transfer learning can be improved by using a better model trained with good (e.g., incremental) source data. We postulate that the proposed approach is feasible and promising and can be further tailored for seamless myoelectric control of powered prosthetics or exoskeletons.

Application of Wireless Virtual Reality Perception and Simulation Technology in Film and Television Animation

Wireless virtual reality integrated multidisciplinary technology, combined with related industries and fields, has changed the way of human-computer interaction and opened up a new field of user experience. In recent years, with the rapid improvement of computer technology and hardware conditions, interactive technology has developed rapidly. The existing wireless virtual reality interactive system is too single and cannot be used in multiple environments. The original system requires a large number of sensor equipment, the cost is high, and the traditional perception technology is too restrictive and cannot realize human-computer interaction more naturally. This paper proposes a dual intention perception algorithm based on the fusion of touch (obtained by experimental simulation equipment), hearing, and vision. The algorithm can perceive the user’s operation intention through the user’s natural behavior and can identify the user’s two intentions at the same time. This paper proposes a navigational interactive mode, which provides users with multimodal intelligent navigation through intelligent perception of user intent and experimental progress. We determine the impact model of the interactive system effect evaluation and analyze its effect evaluation strategy in depth and then further quantify the indicators under the four effect dimensions of information perception, artistic reflection, social entertainment, and aesthetic experience. A combination of qualitative and quantitative methods was used to carry out relevant research on effect evaluation, usability test, and questionnaire interview. The experimental results show that this interactive system has better entertainment effects than other forms of film and television animation, but still needs to pay attention to and strengthen the construction and embodiment of film and television animation content, as well as the optimization and perfection of the fault-tolerant mechanism in the design process.

Gauss: program synthesis by reasoning over graphs

While input-output examples are a natural form of specification for program synthesis engines, they can be imprecise for domains such as table transformations. In this paper, we investigate how extracting readily-available information about the user intent behind these input-output examples helps speed up synthesis and reduce overfitting. We present Gauss, a synthesis algorithm for table transformations that accepts partial input-output examples, along with user intent graphs. Gauss includes a novel conflict-resolution reasoning algorithm over graphs that enables it to learn from mistakes made during the search and use that knowledge to explore the space of programs even faster. It also ensures the final program is consistent with the user intent specification, reducing overfitting. We implement Gauss for the domain of table transformations (supporting Pandas and R), and compare it to three state-of-the-art synthesizers accepting only input-output examples. We find that it is able to reduce the search space by 56×, 73× and 664× on average, resulting in 7×, 26× and 7× speedups in synthesis times on average, respectively.

Intent inference in shared-control teleoperation system in consideration of user behavior

In shared-control teleoperation, rather than directly executing a user’s input, a robot system assists the user via part of autonomy to reduce user’s workload and improve efficiency. Effective assistance is challenging task as it requires correctly inferring the user intent, including predicting the user goal from all possible candidates as well as inferring the user preferred movement in the next step. In this paper, we present a probabilistic formulation for inferring the user intent by taking consideration of user behavior. In our approach, the user behavior is learned from demonstrations, which is then incorporated in goal prediction and path planning. Using maximum entropy principle, two goal prediction methods are tailored according to the similarity metrics between user’s short-term movements and the learned user behavior. We have validated the proposed approaches with a user study—examining the performance of our goal prediction methods in approaching tasks in multiple goals scenario. The results show that our approaches perform well in user goal prediction and are able to respond quickly to dynamic changing of the user’s goals. Comparison analysis shows that the proposed approaches outperform the existing methods especially in scenarios with goal ambiguity.

Real-time decoding of 5 finger movements from 2 EMG channels for mixed reality human-computer interaction

The search for optimized forms of human-computer interaction (HCI) has intensified alongside the growing potential for the combination of biosignals with virtual reality (VR) and augmented reality (AR) to enable the next generation of personal computing. At the core, this requires decoding the user's biosignals into digital commands. Electromyography (EMG) is a biosensor of particular interest due to the ease of data collection, the relatively high signal-to-noise-ratio, its non-invasiveness, and the ability to interpret the signal as being generated by (intentional) muscle activity. Here, we investigate the potential of using data taken from a simple 2-channel EMG setup to differentiate 5 distinct movements. In particular, EMG was recorded from two bipolar sensors over small hand muscles (extensor digitorum, flexor digitorum profundus) while a subject performed 50 trials of dorsal extension and return for each of the five digits. The maximum and the mean data values across the trial were determined for each channel and used as features. A k-nearest neighbors (kNN) classification was performed and overall 5-class classification accuracy reached 94% when using the full trial's time window, while simulated real-time classification reached 90.4% accuracy when using the constructed kNN model (k=3) with a 280ms sliding window. Additionally, unsupervised learning was performed and a homogeneity of 85% was achieved. This study demonstrates that reliable decoding of different natural movements is possible with fewer than one channel per class, even without taking into account temporal features of the signal. The technical feasibility of this approach in a real-time setting was validated by sending real-time EMG data to a custom Unity3D VR application through a Lab Streaming Layer to control a user interface. Further use-cases of gamification and rehabilitation were also examined alongside integration of eye-tracking and gesture recognition for a sensor fusion approach to HCI and user intent.

Question Classification for Efficient QA System

Natural Language Processing (NLP), a subfield of Artificial Intelligence (AI), supports the machine to understand and manipulate the human languages in different sectors. Subsequently, the Question and answering scheme using Machine learning is a challengeable task. For an efficient QA system, understanding the category of a question plays a pivot role in extracting suitable answer. Computers can answer questions requiring single, verifiable answers but fail to answer subjective question demanding deeper understanding of question. Subjective questions can take different forms entailing deeper, multidimensional understanding of context. Identifying the intent of the question helps to extract expected answer from a given passage. Pretrained language models (LMs) have demonstrated excellent results on many language tasks. The paper proposes model of deep learning architecture in hierarchical pattern to learn the semantic of question and extracting appropriate answer. The proposed method converts the given context to fine grained embedding to capture semantic and positional representation, identifies user intent and employs a encoder model to concentrate on answer span. The proposed methods show a remarkable improvement over existing system