Abstract

We investigate a growing body of work that seeks to improve recommender systems through the use of review text. Generally, these papers argue that since reviews ‘explain’ users’ opinions, they ought to be useful to infer the underlying dimensions that predict ratings or purchases. Schemes to incorporate reviews range from simple regularizers to neural network approaches. Our initial findings reveal several discrepancies in reported results, partly due to (e.g.) copying results across papers despite changes in experimental settings or data pre-processing. First, we attempt a comprehensive analysis to resolve these ambiguities. Further investigation calls for discussion on a much larger problem about the “importance" of user reviews for recommendation. Through a wide range of experiments, we observe several cases where state-of-the-art methods fail to outperform existing baselines, especially as we deviate from a few narrowly-defined settings where reviews are useful. We conclude by providing hypotheses for our observations, that seek to characterize under what conditions reviews are likely to be helpful. Through this work, we aim to evaluate the direction in which the field is progressing and encourage robust empirical evaluation.

Abstract

We investigate a growing body of work that seeks to improve recommender systems through the use of review text. Generally, these papers argue that since reviews ‘explain’ users’ opinions, they ought to be useful to infer the underlying dimensions that predict ratings or purchases. Schemes to incorporate reviews range from simple regularizers to neural network approaches. Our initial findings reveal several discrepancies in reported results, partly due to (e.g.) copying results across papers despite changes in experimental settings or data pre-processing. First, we attempt a comprehensive analysis to resolve these ambiguities. Further investigation calls for discussion on a much larger problem about the “importance" of user reviews for recommendation. Through a wide range of experiments, we observe several cases where state-of-the-art methods fail to outperform existing baselines, especially as we deviate from a few narrowly-defined settings where reviews are useful. We conclude by providing hypotheses for our observations, that seek to characterize under what conditions reviews are likely to be helpful. Through this work, we aim to evaluate the direction in which the field is progressing and encourage robust empirical evaluation.

Abstract

Learning effective contextual-bandit policies from past actions of a deployed system is highly desirable in many settings (e.g. voice assistants, recommendation, search), since it enables the reuse of large amounts of log data. State-of-the-art methods for such offpolicy learning, however, are based on inverse propensity score (IPS) weighting. A key theoretical requirement of IPS weighting is that the policy that logged the data has "full support", which typically translates into requiring non-zero probability for any action in any context. Unfortunately, many real-world systems produce support deficient data, especially when the action space is large, and we show how existing methods can fail catastrophically. To overcome this gap between theory and applications, we identify three approaches that provide various guarantees for IPS-based learning despite the inherent limitations of support-deficient data: restricting the action space, reward extrapolation, and restricting the policy space. We systematically analyze the statistical and computational properties of these three approaches, and we empirically evaluate their effectiveness. In addition to providing the first systematic analysis of support-deficiency in contextual-bandit learning, we conclude with recommendations that provide practical guidance.

Abstract

Identifying types of online users’ toxic behavior reveals important insights from social media interactions, including whether a user be- comes “radicalized” (more toxic) or “pacified” (less toxic) over time. In this research, we design two metrics to identify toxic user types: F score that captures the changes in a user’s toxicity, and G score that captures the direction of the shift taking place in the user’s toxicity pattern. We apply these metrics to a dataset of 4M user comments from Reddit by defining four toxic user types based on the toxicity scores of a user’s comments: (a) Steady Users whose toxicity scores are steady over time, (b) Fickle-Minded Users that switch between toxic and non-toxic commenting, (c) Pacified Users whose com- menting becomes less toxic in time, and (d) Radicalized Users that become gradually toxic. Findings from the Reddit dataset indicate that fickle-minded users form the largest group (31.2%), followed by pacified (25.8%), radicalized (25.4%), and steadily toxic users (17.6%). The results suggest that the most typical behavior type of toxicity is switching between toxic and non-toxic commenting. This research has implications for preserving the user-friendliness of online communities by identifying continuously toxic users and users in danger of becoming radicalized (in terms of their toxic behavior), and designing interventions to mitigate these behavior types. Using the metrics we have defined, identifying these user types becomes possible. More research is needed to understand why these patterns take place and how they could be mitigated.

Abstract

The paper focuses on the classification of normal and abnormal heart sounds through a low-cost, high-performance PhonoCardioGraphy system using MEMS microphones with good sensitivity for the heart sounds. The proposed algorithm classifies the heart sound components using wavelet decomposition and Shannon energy. The PhonoCardioGraphy system, with the proposed algorithm tested by taking 1756 results, includes both normal and abnormal heart sounds. Experimental results revealed a good sensitivity of 99.17% and detection error rate of 1.5%.

Abstract

Given an image or a video captured from a monocular camera, amodal layout estimation is the task of predicting semantics and occupancy in bird's eye view. The term amodal implies we also reason about entities in the scene that are occluded or truncated in image space. While several recent efforts have tackled this problem, there is a lack of standardization in task specification, datasets, and evaluation protocols. We address these gaps with AutoLay, a dataset and benchmark for amodal layout estimation from monocular images. AutoLay encompasses driving imagery from two popular datasets: KITTI [1] and Argoverse [2]. In addition to fine-grained attributes such as lanes, sidewalks, and vehicles, we also provide semantically annotated 3D point clouds. We implement several baselines and bleeding edge approaches, and release our data and code

Abstract

With the growing use of social media and its availability, many instances of the use of offensive language have been observed across multiple languages and domains. This phenomenon has given rise to the growing need to detect the offensive language used in social media crosslingually. In OffensEval 2020, the organizers have released the multilingual Offensive Language Identification Dataset (mOLID), which contains tweets in five different languages, to detect offensive language. In this work, we introduce a cross-lingual inductive approach to identify the offensive language in tweets using the contextual word embedding XLM-RoBERTa (XLM-R). We show that our model performs competitively on all five languages, obtaining the fourth position in the English task with an F1-score of 0.919 and eighth position in the Turkish task with an F1-score of 0.781. Further experimentation proves that our model works competitively in a zero-shot learning environment, and is extensible to other languages.

Abstract

Internet of Things (IoT) is an emerging technology, which integrates all everyday objects and services into the com- mon network platform like Internet. With the help of IoT, every physical object is converted as smart object. Each of the physical object is identified uniquely and can sense the information in the environment and to take decisions based on that. IoT is useful in many of the applications, including home automation, retail, industry control, agriculture, health monitoring, etc. In healthcare domain, some of the IoT-based applications include, monitoring old people for fall detection, monitoring temperature conditions inside refrigerator for storing medicines, vaccines, and monitoring of patient’s condition in hospitals and home. Machine Learning applications in Healthcare also seems to be endless. In this paper, we have investigated IoT based applications in healthcare and performed an experimental analysis of he

Abstract

In recent years, the increased demand for wireless multimedia services has led to huge bandwidth crisis. The concept of Cognitive Radio Networks (CRN) has been proposed as a solution. In CRN, higher Primary User Activity leads to an increased number of Pre-empted Secondary Users (PSU). Since the PSUs are given priority over New Secondary Users (NSU), the NSUs might get starved. Due to this priority of PSUs less number low priority users can be served. The advent of the Internet of Things, the traffic pattern has become unpredictable. All this reasons required a Dynamic Cognitive Channel Allocation (DCCA) algorithm which has been proposed in this work which balances the service among the PSUs and NSUs as per priority of SU as well as avoiding starvation. Analysis of the proposed algorithm using Continuous Time Markov Chain Model shows an improvement over existing literature of 44.44% in NSU dropping probability and 22.8% improvement in NSU blocking probability with respect to arrival of PUs. Varying the arrival of SUs has resulted in an improvement of 58% and 45.75% in the dropping and blocking probabilities of NSUs respectively. The starvation of SUs has lowered by 32.52% with varying arrival rate of PU and 41.38% with varying arrival rate of SUs

Abstract

Unmanned Aerial Vehicle (UAV)-enabled mobile edge computing (MEC) is considered to offer computational capabilities to the resource-constraints end-users. In this paper, we study the task offloading strategy in UAV-enabled MEC systems, where end-users offload the computation-intensive tasks to the UAV to minimize the overall cost in terms of the weighted delay and energy consumption. The end-users either process the task by itself or offload the tasks to the UAV that acts as a computing access point. However, due to the computation bottleneck and limited channel capacity between UAV and the end-users, it becomes a challenging issue to offload the entire tasks to the UAV. Thus, to find the optimal offloading decision for the tasks generated by the end-users, we build a distributed deep neural network (DNN). In the proposed distributed DNN model, we train multiple DNNs in the same training instance, and finally, for validation, we select the DNN that gives the least training loss. For faster convergence of the training process, we use the optimal generated offloading decision using a Quadratically Constrained Linear Program (QCLP) with Semidefinite Relaxation (SDR). The extensive simulation results show that the offloading decision produced by the trained DNN can achieve near-optimal performance with numerous system parameter settings.