An intent detector is a central component of any task-oriented conversational system. The goal of the intent detector is to identify the user’s goal by classifying natural language utterances. In recent years, research has focused on supervised intent detection models. Supervised learning approaches cannot accommodate unseen intents, which may emerge after the system has been deployed– the more practically relevant setting, known as zero-shot intent detection. The existing zero-shot learning approaches split a dataset into seen and unseen intents for training and evaluations without taking the sensitivity of the data collection process into account. That is, humans tend to use repeated vocabulary and compose sentences with similar compositional structures. We argue that the source-to-target relationship learning objective of zero-shot approaches under typical data split procedure renders the zero-shot models prone to misclassifications when target intents are divergent from source intents. To this end, we propose INTEND , a zero-shot INTENt Detection methodology that leverages contrastive transfer learning and employs a zero-shot learning paradigm in its true sense. First, in contrast to partitioning the training and testing sets from the same dataset, we demonstrate that selecting training and testing sets from two different datasets allows for rigorous zero-shot intent detection evaluations. Second, our employed contrastive learning goal is model-agnostic and encourages the system to focus on learning a generic similarity function, rather than on commonly encountered patterns in the training set. We conduct extensive experimental evaluations using a range of transformer models on four public intent detection datasets for up to 150 unseen classes. Our experimental results show that INTEND consistently outperforms state-of-the-art zero-shot techniques by a substantial margin. Furthermore, our approach achieves significantly better performance than few-shot intent detection models.

Task-oriented dialog systems empower users to accomplish their goals by facilitating intuitive and expressive natural language interactions. State-of-the-art approaches in task-oriented dialog systems formulate the problem as a conditional sequence generation task and fine-tune pre-trained causal language models in the supervised setting. This requires labeled training data for each new domain or task, and acquiring such data is prohibitively laborious and expensive, thus making it a bottleneck for scaling systems to a wide range of domains. To overcome this challenge, we introduce a novel Zero-Shot generalizable end-to-end Task-oriented Dialog system, ZS-ToD, that leverages domain schemas to allow for robust generalization to unseen domains and exploits effective summarization of the dialog history. We employ GPT-2 as a backbone model and introduce a two-step training process where the goal of the first step is to learn the general structure of the dialog data and the second step optimizes the response generation as well as intermediate outputs, such as dialog state and system actions. As opposed to state-of-the-art systems that are trained to fulfill certain intents in the given domains and memorize task-specific conversational patterns, ZS-ToD learns generic task-completion skills by comprehending domain semantics via domain schemas and generalizing to unseen domains seamlessly. We conduct an extensive experimental evaluation on SGD and SGD-X datasets that span up to 20 unique domains and ZS-ToD outperforms state-of-the-art systems on key metrics, with an improvement of +17% on joint goal accuracy and +5 on inform. Additionally, we present a detailed ablation study to demonstrate the effectiveness of the proposed components and training mechanism

Objective:To evaluate whether the deep learning (DL) segmentation methods from the six teams that participated in the IWOAI 2019 Knee Cartilage Segmentation Challenge are appropriate for quantifying cartilage loss in longitudinal clinical trials. Design:We included 556 subjects from the Osteoarthritis Initiative study with manually read cartilage volume scores for the baseline and 1-year visits. The teams used their methods originally trained for the IWOAI 2019 challenge to segment the 1130 knee MRIs. These scans were anonymized and the teams were blinded to any subject or visit identifiers. Two teams also submitted updated methods. The resulting 9,040 segmentations are available online.The segmentations included tibial, femoral, and patellar compartments. In post-processing, we extracted medial and lateral tibial compartments and geometrically defined central medial and lateral femoral sub-compartments. The primary study outcome was the sensitivity to measure cartilage loss as defined by the standardized response mean (SRM). Results:For the tibial compartments, several of the DL segmentation methods had SRMs similar to the gold standard manual method. The highest DL SRM was for the lateral tibial compartment at 0.38 (the gold standard had 0.34). For the femoral compartments, the gold standard had higher SRMs than the automatic methods at 0.31/0.30 for medial/lateral compartments. Conclusion:The lower SRMs for the DL methods in the femoral compartments at 0.2 were possibly due to the simple sub-compartment extraction done during post-processing. The study demonstrated that state-of-the-art DL segmentation methods may be used in standardized longitudinal single-scanner clinical trials for well-defined cartilage compartments.

Recommender systems have become ubiquitous in our digital lives, from recommending products on e-commerce websites to suggesting movies and music on streaming platforms. Existing recommendation datasets, such as Amazon Product Reviews and MovieLens, greatly facilitated the research and development of recommender systems in their respective domains. While the number of mobile users and applications (aka apps) has increased exponentially over the past decade, research in mobile app recommender systems has been significantly constrained, primarily due to the lack of high-quality benchmark datasets, as opposed to recommendations for products, movies, and news. To facilitate research for app recommendation systems, we introduce a large-scale dataset, called MobileRec. We constructed MobileRec from users' activity on the Google play store. MobileRec contains 19.3 million user interactions (i.e., user reviews on apps) with over 10K unique apps across 48 categories. MobileRec records the sequential activity of a total of 0.7 million distinct users. Each of these users has interacted with no fewer than five distinct apps, which stands in contrast to previous datasets on mobile apps that recorded only a single interaction per user. Furthermore, MobileRec presents users' ratings as well as sentiments on installed apps, and each app contains rich metadata such as app name, category, description, and overall rating, among others. We demonstrate that MobileRec can serve as an excellent testbed for app recommendation through a comparative study of several state-of-the-art recommendation approaches. The MobileRec dataset is available at https://huggingface.co/datasets/recmeapp/mobilerec.

Intent detector is a central component of any task-oriented conversational system. The goal of the intent detector is to identify the user's goal by classifying natural language utterances. In recent years, research has focused on supervised intent detection models. Supervised learning approaches cannot accommodate unseen intents, which may emerge after the system has been deployed — the more practically relevant setting, known as zero-shot intent detection. The existing zero-shot learning approaches split a dataset into seen and unseen intents for training and evaluations without taking the sensitivity of the data collection process into account. That is, humans tend to use repeated vocabulary and compose sentences with similar compositional structures. We argue that the source-to-target relationship learning objective of zero-shot approaches under typical data split procedure renders the zero-shot models prone to misclassifications when target intents are divergent from source intents. To this end, we propose INTEND, a zero-shot INTENt Detection methodology that leverages contrastive transfer learning and employs a zero-shot learning paradigm in its true sense. First, in contrast to partitioning the training and testing sets from the same dataset, we demonstrate that selecting training and testing sets from two different datasets, allows for rigorous zero-shot intent detection evaluations. Second, our employed contrastive learning goal encourages the system to focus on learning a generic similarity function, rather than on commonly encountered patterns in the training set. We conduct extensive experimental evaluations using four public intent detection datasets for up to 150 unseen classes. Our experimental results show that INTEND consistently outperforms state-of-the-art zero-shot techniques by a substantial margin. Furthermore, our approach achieves significantly better performance than few-shot intent detection models.

Most of the prior work on goal-oriented dialog systems has concentrated on developing systems that heavily rely on the relevant domain APIs to generate a response. However, in the real world, users frequently make such requests that the provided APIs cannot handle, we call them “off-script” queries. Ideally, existing information retrieval approaches could have leveraged relevant enterprise's unstructured data sources to retrieve the appropriate information to synthesize responses for such queries. But, in multi-turn dialogs, these queries oftentimes are not self-contained, rendering most of the existing information retrieval methods ineffective, and the dialog systems end up responding “sorry I don't know this”. That is, off-script queries may mention entities from the previous dialog turns (often expressed through pronouns) or do not mention the referred entities at all. These two problems are known as coreference resolution and ellipsis, respectively; extensively studied research problems in the supervised settings. In this paper, we first build a dataset of off-script and contextual user queries for goal-oriented dialog systems. Then, we propose a zero-label approach to rewrite the contextual query as a self-contained one by leveraging the dialog's state. We propose two parallel coreference and ellipsis resolution pipelines to synthesize candidate queries, rank and select the candidates based on the pre-trained language model GPT-2, and refine the selected self-contained query with the pre-trained BERT. We show that our approach leads to higher quality expanded questions compared to state-of-the-art supervised methods, on our dataset and existing datasets. The key advantage of our novel zero-label approach is that it requires no labeled training data and can be applied to any domain seamlessly, in contrast to previous work that requires labeled training data for each new domain.

Purpose: To evaluate the performance of 8 state-of-the-art knee MRI segmentations methods for assessment of cartilage quantity changes in clinical trials.While deep learning (DL) methods have demonstrated impressive, near-human performance for many image analysis tasks, some of the methods have been shown to be sensitive to changes in data distributions. For a multi-center, longitudinal study, there will be distribution shifts between centers and over time due to scanner software (e.g. image reconstruction methods) and hardware upgrades (coils, gradients, etc.).

Task-oriented dialog systems enable users to accomplish tasks using natural language. State-of-the-art systems respond to users in the same way regardless of their personalities, although personalizing dialogues can lead to higher levels of adoption and better user experiences. Building personalized dialog systems is an important, yet challenging endeavor, and only a handful of works took on the challenge. Most existing works rely on supervised learning approaches and require laborious and expensive labeled training data for each user profile. Additionally, collecting and labeling data for each user profile is virtually impossible. In this work, we propose a novel framework, P-ToD, to personalize task-oriented dialog systems capable of adapting to a wide range of user profiles in an unsupervised fashion using a zero-shot generalizable reward function. P-ToD uses a pre-trained GPT-2 as a backbone model and works in three phases. Phase one performs task-specific training. Phase two kicks off unsupervised personalization by leveraging the proximal policy optimization algorithm that performs policy gradients guided by the zero-shot generalizable reward function. Our novel reward function can quantify the quality of the generated responses even for unseen profiles. The optional final phase fine-tunes the personalized model using a few labeled training examples. We conduct extensive experimental analysis using the personalized bAbI dialogue benchmark for five tasks and up to 180 diverse user profiles. The experimental results demonstrate that P-ToD, even when it had access to zero labeled examples, outperforms state-of-the-art supervised personalization models and achieves competitive performance on BLEU and ROUGE metrics when compared to a strong fully-supervised GPT-2 baseline.

Skeleton-based action recognition has recently attracted a lot of attention. Researchers are coming up with new approaches for extracting spatio-temporal relations and making considerable progress on large-scale skeleton-based datasets. Most of the architectures being proposed are based upon recurrent neural networks (RNNs), convolutional neural networks (CNNs) and graph-based CNNs. When it comes to skeleton-based action recognition, the importance of long term contextual information is central which is not captured by the current architectures. In order to come up with a better representation and capturing of long term spatio-temporal relationships, we propose three variants of Self-Attention Network (SAN), namely, SAN-V1, SAN-V2 and SAN-V3. Our SAN variants has the impressive capability of extracting high-level semantics by capturing long-range correlations. We have also integrated the Temporal Segment Network (TSN) with our SAN variants which resulted in improved overall performance. Different configurations of Self-Attention Network (SAN) variants and Temporal Segment Network (TSN) are explored with extensive experiments. Our chosen configuration outperforms state-of-the-art Top-1 and Top-5 by 4.4% and 7.9% respectively on Kinetics and shows consistently better performance than state-of-the-art methods on NTU RGB+D.

Wireless multimedia sensor network has been renewed to the distributed model by incorporating the single board computer (i.e., camera nodes and mobile beacons). The redesigned camera nodes, such as unmanned aerial vehicle and mobile beacon nodes are limited in terms of memory, computation power, bandwidth, and node power while the real-time processing requires the multimedia contents transmission over the wireless interface in the presence of all these constraints. To tackle this problem, advanced video compression schemes have been adopted by which bit rate overhead over the network is greatly improved. H.264/moving picture experts group (MPEG)-4 is a state-of-the-art video compression standard developed jointly by International Telecommunication Union Telecommunication Standardization Sector video coding experts group and ISO/Joint Technical Committee of International Electrotechnical Commission MPEG. The H.264 provides a number of tunable parameters (e.g., quantization parameter (QP) and number of reference frames (NoRF)) for tailoring the encoding process as per requirements. In this paper, we propose a framework consisting of a joint parameter cost function to find the best trade-off between QP and NoRF with the policy of controlling the bit-rate (BR) while maintaining reasonably good video quality without compromising on processing delay (PD) and nodal power consumption (NPC). The framework has been proposed by transforming the multi-objective problem into a convex optimization problem. It is shown that an optimal trade-off between these conflicting objectives can be achieved by choosing appropriate values for QP and NoRF. Results for optimal values of QP and NoRF with corresponding trade-off between BR, quality of multimedia contents, low PD and low NPC are presented.

In this paper, we propose an efficient self-calibration mechanism, achieving localization as well as orientation, for wireless acoustic sensor network. Time difference of arrival (TDOA) based on wireless and acoustic signals, while direction of arrival (DOA) based on acoustic only, are used jointly to achieve acoustic sensor node calibration. Each sensor node is equipped with three microphones and a wireless transceiver. Sensor nodes are calibrated by using two beacon positions with respect to a reference node whose position is assumed to be known. Our approach requires a single moving beacon (MB) equipped with RF and acoustic signal sources. Once a node has been calibrated (in terms of location and orientation) it can be considered as a valid reference for the remaining uncalibrated nodes. We have developed hardware platform to validate the proposed calibration mechanism. Performance results show the effectiveness and usefulness of the proposed mechanism.