This paper describes LeVoice automatic speech recognition systems to track2 of intelligent cockpit speech recognition challenge 2022. Track2 is a speech recognition task without limits on the scope of model size. Our main points include deep learning based speech enhancement, text-to-speech based speech generation, training data augmentation via various techniques and speech recognition model fusion. We compared and fused the hybrid architecture and two kinds of end-to-end architecture. For end-to-end modeling, we used models based on connectionist temporal classification/attention-based encoder-decoder architecture and recurrent neural network transducer/attention-based encoder-decoder architecture. The performance of these models is evaluated with an additional language model to improve word error rates. As a result, our system achieved 10.2\% character error rate on the challenge test set data and ranked third place among the submitted systems in the challenge.
Deep learning has seen increasing applications in time series in recent years. For time series anomaly detection scenarios, such as in finance, Internet of Things, data center operations, etc., time series usually show very flexible baselines depending on various external factors. Anomalies unveil themselves by lying far away from the baseline. However, the detection is not always easy due to some challenges including baseline shifting, lacking of labels, noise interference, real time detection in streaming data, result interpretability, etc. In this paper, we develop a novel deep architecture to properly extract the baseline from time series, namely Deep Baseline Network (DBLN). By using this deep network, we can easily locate the baseline position and then provide reliable and interpretable anomaly detection result. Empirical evaluation on both synthetic and public real-world datasets shows that our purely unsupervised algorithm achieves superior performance compared with state-of-art methods and has good practical applications.
This paper aims to interpret how deepfake detection models learn artifact features of images when just supervised by binary labels. To this end, three hypotheses from the perspective of image matching are proposed as follows. 1. Deepfake detection models indicate real/fake images based on visual concepts that are neither source-relevant nor target-relevant, that is, considering such visual concepts as artifact-relevant. 2. Besides the supervision of binary labels, deepfake detection models implicitly learn artifact-relevant visual concepts through the FST-Matching (i.e. the matching fake, source, target images) in the training set. 3. Implicitly learned artifact visual concepts through the FST-Matching in the raw training set are vulnerable to video compression. In experiments, the above hypotheses are verified among various DNNs. Furthermore, based on this understanding, we propose the FST-Matching Deepfake Detection Model to boost the performance of forgery detection on compressed videos. Experiment results show that our method achieves great performance, especially on highly-compressed (e.g. c40) videos.
With the growing popularity of deep-learning models, model understanding becomes more important. Much effort has been devoted to demystify deep neural networks for better interpretability. Some feature attribution methods have shown promising results in computer vision, especially the gradient-based methods where effectively smoothing the gradients with reference data is key to a robust and faithful result. However, direct application of these gradient-based methods to NLP tasks is not trivial due to the fact that the input consists of discrete tokens and the "reference" tokens are not explicitly defined. In this work, we propose Locally Aggregated Feature Attribution (LAFA), a novel gradient-based feature attribution method for NLP models. Instead of relying on obscure reference tokens, it smooths gradients by aggregating similar reference texts derived from language model embeddings. For evaluation purpose, we also design experiments on different NLP tasks including Entity Recognition and Sentiment Analysis on public datasets as well as key feature detection on a constructed Amazon catalogue dataset. The superior performance of the proposed method is demonstrated through experiments.
Natural language spatial video grounding aims to detect the relevant objects in video frames with descriptive sentences as the query. In spite of the great advances, most existing methods rely on dense video frame annotations, which require a tremendous amount of human effort. To achieve effective grounding under a limited annotation budget, we investigate one-shot video grounding, and learn to ground natural language in all video frames with solely one frame labeled, in an end-to-end manner. One major challenge of end-to-end one-shot video grounding is the existence of videos frames that are either irrelevant to the language query or the labeled frames. Another challenge relates to the limited supervision, which might result in ineffective representation learning. To address these challenges, we designed an end-to-end model via Information Tree for One-Shot video grounding (IT-OS). Its key module, the information tree, can eliminate the interference of irrelevant frames based on branch search and branch cropping techniques. In addition, several self-supervised tasks are proposed based on the information tree to improve the representation learning under insufficient labeling. Experiments on the benchmark dataset demonstrate the effectiveness of our model.
This paper reports the results and post-challenge analyses of ChaLearn's AutoDL challenge series, which helped sorting out a profusion of AutoML solutions for Deep Learning (DL) that had been introduced in a variety of settings, but lacked fair comparisons. All input data modalities (time series, images, videos, text, tabular) were formatted as tensors and all tasks were multi-label classification problems. Code submissions were executed on hidden tasks, with limited time and computational resources, pushing solutions that get results quickly. In this setting, DL methods dominated, though popular Neural Architecture Search (NAS) was impractical. Solutions relied on fine-tuned pre-trained networks, with architectures matching data modality. Post-challenge tests did not reveal improvements beyond the imposed time limit. While no component is particularly original or novel, a high level modular organization emerged featuring a "meta-learner", "data ingestor", "model selector", "model/learner", and "evaluator". This modularity enabled ablation studies, which revealed the importance of (off-platform) meta-learning, ensembling, and efficient data management. Experiments on heterogeneous module combinations further confirm the (local) optimality of the winning solutions. Our challenge legacy includes an ever-lasting benchmark (http://autodl.chalearn.org), the open-sourced code of the winners, and a free "AutoDL self-service".
A great surge of the global navigation satellite system (GNSS) development excavates the potential of promoting pomposity in many state-of-art technologies, e.g., autonomous ground vehicles (AGVs). Nevertheless, the GNSS is fragile to the various ground interferences which significantly break down the continuity of the navigation system. Meanwhile, the GNSS-based next-generation navigation devices are being developed to be smaller, more low-cost, and lightweight as forecasted by the commercial market. This work aims to answer the question of whether the smartphone inertial measurement unit (IMU) is sufficient to support the GNSS baseband. Thus, a cascaded ultra-tightly integrated GNSS/inertial navigation system (INS) technique, where the consumer-level smartphone sensors are used, is proposed to improve the baseband performance of GNSS software-defined radios (SDRs). To integrate the GNSS baseband, a Doppler value is predicted based on an integrated extended Kalman filter (EKF) navigator where the pseudo-range-state-based measurements of GNSS and INS are fused, and it is used to assist the numerically controlled oscillator (NCO) algorithms. Then, an ultra-tight integration platform is built with an upgraded GNSS SDR of which baseband processing is integrated with the INS mechanization algorithm. Finally, by comparing with the previous algorithms, both tracking-level and carrier-based positioning performances are assessed in the proposed platform for the smartphone-IMU-aided GNSS baseband via kinematic AGV field tests. The experimental results demonstrate the performance of the tracking ability and the high-precision positioning of the proposed ultra-tight integration algorithms using the smartphone IMU.
Federated learning (FL) is a privacy-preserving machine learning paradigm that enables collaborative training among geographically distributed and heterogeneous users without gathering their data. Extending FL beyond the conventional supervised learning paradigm, federated Reinforcement Learning (RL) was proposed to handle sequential decision-making problems for various privacy-sensitive applications such as autonomous driving. However, the existing federated RL algorithms directly combine model-free RL with FL, and thus generally have high sample complexity and lack theoretical guarantees. To address the above challenges, we propose a new federated RL algorithm that incorporates model-based RL and ensemble knowledge distillation into FL. Specifically, we utilise FL and knowledge distillation to create an ensemble of dynamics models from clients, and then train the policy by solely using the ensemble model without interacting with the real environment. Furthermore, we theoretically prove that the monotonic improvement of the proposed algorithm is guaranteed. Extensive experimental results demonstrate that our algorithm obtains significantly higher sample efficiency compared to federated model-free RL algorithms in the challenging continuous control benchmark environments. The results also show the impact of non-IID client data and local update steps on the performance of federated RL, validating the insights obtained from our theoretical analysis.
Temporal relational data, perhaps the most commonly used data type in industrial machine learning applications, needs labor-intensive feature engineering and data analyzing for giving precise model predictions. An automatic machine learning framework is needed to ease the manual efforts in fine-tuning the models so that the experts can focus more on other problems that really need humans' engagement such as problem definition, deployment, and business services. However, there are three main challenges for building automatic solutions for temporal relational data: 1) how to effectively and automatically mining useful information from the multiple tables and the relations from them? 2) how to be self-adjustable to control the time and memory consumption within a certain budget? and 3) how to give generic solutions to a wide range of tasks? In this work, we propose our solution that successfully addresses the above issues in an end-to-end automatic way. The proposed framework, AutoSmart, is the winning solution to the KDD Cup 2019 of the AutoML Track, which is one of the largest AutoML competition to date (860 teams with around 4,955 submissions). The framework includes automatic data processing, table merging, feature engineering, and model tuning, with a time\&memory controller for efficiently and automatically formulating the models. The proposed framework outperforms the baseline solution significantly on several datasets in various domains.