Deep Learning Processing Research Articles

高等教育领域中ChatGPT的应用风险及其应对研究

ChatGPT technology, based on deep learning and natural language processing, excels in generating fluid, coherent, and meaningful text dialogues. Its integration into higher education has ushered in transformative changes with positive impacts. The widespread adoption of ChatGPT in education empowers college students to pursue autonomous learning, while at the teaching level, it enables collaborative human-machine interactions, potentially creating exceptional educators. Additionally, ChatGPT enhances teaching management and evaluation, significantly boosting their efficiency. However, its application also introduces inherent risks, including challenges in ensuring the legality of its data sources, potential student information leaks, risks of academic misconduct due to ChatGPT misuse, and the potential diminishment of the teacher's role through excessive interaction. This article offers response suggestions to address these concerns: firstly, ensuring the legality of ChatGPT data sources in compliance with the law; secondly, establishing legal frameworks to define ChatGPT product copyrights; thirdly, fostering a balanced and ethical human-computer interaction; fourthly, reinforcing ethical standards governing ChatGPT's integration into higher education.

A deep implicit memory Gaussian network for time series forecasting

In recent years, significant achievements have been made in time series forecasting using deep learning methods, particularly the Long Short-Term Memory Network (LSTM). However, time series often exhibit complex patterns and relationships like trends, seasonal patterns and irregularities, and LSTM networks fail to effectively strengthen long-term dependence in time series data, which may lead to inaccurate forecasting effect or performance degradation. Therefore, building a model to explore the temporal dependence within the time series data completely still remains a challenge. In this paper, we propose a novel Deep Implicit Memory Gaussian (DIMG) Network based on bidirectional deep memory kernel process and the implicit features enhancement method for time series forecasting. We first use the implicit features enhancement method to obtain the hidden features of the data according to the nonlinear mapping characteristics of encoder. Then, a new deep learning process called bidirectional deep memory kernel has been developed, which merges the structural properties of deep learning with the adaptability of kernel methods to capture intricate information and memory structures in sequential data. This process fully encapsulates the structure of Bi-LSTM and Gaussian process regression (GPR). Finally, the performance of the proposed model is evaluated on two real-world datasets. The experimental results verify that our model outperforms other reported methods in terms of prediction accuracy.

A new method based on deep learning and image processing for detection of strabismus with the Hirschberg test

Strabismus is a condition in which one or both eyes do not work in parallel or in harmony. People with strabismus have one eye looking straight ahead while the other eye looks inwards, outwards, upwards or downwards. This condition can affect both eyes. Strabismus is a common eye condition that affects about 4 % of the world's population. Tests such as Hirschberg, Cover and Krimsky are used to detect strabismus. In the Hirschberg test, a light source is held at a distance of 50 cm so that it falls on the centre of each eye. The horizontal and vertical distance between the centre of gravity of the light reflected from the cornea and the centre of the pupil indicates the degree of strabismus. In this study, deep learning and image processing algorithms are used to detect the eye, corneal reflection, iris and pupil on a patient's facial image. Based on the Hirschberg test, the horizontal and vertical shifts for both eyes were measured to determine the patient's degree of strabismus. In this way, the Hirschberg test used in strabismus screening was performed automatically by software. The correct detection of the pupil and the light reflected from the cornea by the algorithm means that the eye has been measured correctly. The software was tested on the facial images of 88 strabismic patients of different sexes and ages. 91 % of the 88 patients, or 80 patients, had their left eye measured correctly. 90 % of the 88 patients, or 79 patients, had their right eye measured correctly. The results for each eye obtained from the correct measurements were found to have an error of maximum ± 2°. This error is due to the fact that a real eye is in three-dimensional space, while the digital eye image is in two-dimensional space, and was only observed in the test results of some patients. This algorithm can be tested on patients of all ages and is not affected by morphological differences in the patients' faces. Successful results have been observed experimentally that this newly proposed method can be used in strabismus screening.

Binary code similarity analysis based on naming function and common vector space

Binary code similarity analysis is widely used in the field of vulnerability search where source code may not be available to detect whether two binary functions are similar or not. Based on deep learning and natural processing techniques, several approaches have been proposed to perform cross-platform binary code similarity analysis using control flow graphs. However, existing schemes suffer from the shortcomings of large differences in instruction syntaxes across different target platforms, inability to align control flow graph nodes, and less introduction of high-level semantics of stability, which pose challenges for identifying similar computations between binary functions of different platforms generated from the same source code. We argue that extracting stable, platform-independent semantics can improve model accuracy, and a cross-platform binary function similarity comparison model N_Match is proposed. The model elevates different platform instructions to the same semantic space to shield their underlying platform instruction differences, uses graph embedding technology to learn the stability semantics of neighbors, extracts high-level knowledge of naming function to alleviate the differences brought about by cross-platform and cross-optimization levels, and combines the stable graph structure as well as the stable, platform-independent API knowledge of naming function to represent the final semantics of functions. The experimental results show that the model accuracy of N_Match outperforms the baseline model in terms of cross-platform, cross-optimization level, and industrial scenarios. In the vulnerability search experiment, N_Match significantly improves hit@N, the mAP exceeds the current graph embedding model by 66%. In addition, we also give several interesting observations from the experiments. The code and model are publicly available at https://www.github.com/CSecurityZhongYuan/Binary-Name_Match.

Retracted: Multifunctional Robot Grasping System Based on Deep Learning and Image Processing

Retracted: Multifunctional Robot Grasping System Based on Deep Learning and Image Processing

VADF: V ersatile A pproximate D ata F ormats for Energy-Efficient Computing

Approximate computing (AC) techniques provide overall performance gains in terms of power and energy savings at the cost of minor loss in application accuracy. For this reason, AC has emerged as a viable method for efficiently supporting several compute-intensive applications, e.g., machine learning, deep learning, and image processing, that can tolerate bounded errors in computations. However, most prior techniques do not consider the possibility of soft errors or malicious bit-flips in AC systems. These errors may interact with approximation-introduced errors in unforeseen ways, leading to disastrous consequences, such as the failure of computing systems. A recent research effort, FTApprox (DATE’21) proposes an error-resilient approximate data format. FTApprox stores two blocks, starting from the one containing the most significant valid (MSV) bit. It also stores location of the MSV block and protects them using error-correcting bits (ECBs). However, FTApprox has crucial limitations such as lack of flexibility, redundantly storing zeros in the MSV, etc. In this paper, we propose a novel storage format named Versatile Approximate Data Format (VADF) for storing approximate integer numbers while providing resilience to soft errors. VADF prescribes rules for storing, for example, a 32-bit number in either 8-bit, 12-bit or 16-bit numbers. VADF identifies the MSV bit and stores a certain number of bits following the MSV bit. It also stores the location of the MSV bit and protects it by ECBs. VADF does not explicitly store the MSB bit itself and this prevents VADF from accruing significant errors. VADF incurs lower error than both truncation methodologies and FTApprox. We further evaluate five image-processing and machine-learning applications and confirm that VADF provides higher application quality than FTApprox in the presence and absence of soft errors. Finally, VADF allows the use of narrow arithmetic units. For example, instead of using a 32-bit multiplier/adder, one can first use VADF (or FTApprox) to compress the data and then use a 8-bit multiplier/adder. Through this approach, VADF facilitates 95.97% and 79.3% energy savings in multiplication and addition, respectively. However, the subsequent re-conversion of the 8-bit output data to 32-bit data using Inv-VADF(16,3,32) diminishes the energy savings by 9.6% for addition and 0.56% for multiplication operation, respectively. The code is available at https://github.com/CandleLabAI/VADF-ApproximateDataFormat-TECS .

A system-on-chip solution for deep learning-based automatic fetal biometric measurement

Ultrasound assessment of fetal biometry is widely used for gestational age prediction and abnormal fetal growth diagnosis. However, manual biometric measurements require time-consuming procedures and current automated measurement approaches remain challenging in terms of accuracy and computational complexity. This paper proposes deep learning-based efficient automatic fetal biometry measurement method for system-on-chip (SoC) solution. For end-to-end automated measurements, a hardware-friendly bilateral segmentation network (H-BiSeNet) was designed and optimized for multiple fetal objects (i.e., head, abdomen and femur), and fetal biometric parameters were then calculated with a robust multiple keypoint detection algorithm. For hardware implementation with a SoC, a deep learning processing unit (DPU) was employed to accelerate the proposed segmentation network, and biometric measurements with image pre- and postprocessing were conducted on an application processing unit (APU). A total of 6000 fetal ultrasound images in the three regions were collected from 2568 subjects, and the dataset was utilized to train the proposed network. In the performance evaluation, the proposed segmentation network with optimal quantization (H-BiSeNet-Q) outperformed other segmentation networks (i.e., U-Net, U-Net++, Attention U-Net, U-Net3+, DeepLabv3 + and BiSeNetv2) in terms of the average Dice coefficient, and automated measurements with H-BiSeNet-Q were highly correlated with manual measurements (r ≥ 0.983, p < 0.01 for all parameters, i.e., biparietal diameter, head circumference, abdominal circumference and femur length). In the execution time analysis, H-BiSeNet-Q ran at 4.25 frames per second (FPS) with parallelism corresponding to 80% resource utilization in the SoC. These results showed feasibility in low-resource hardware settings such as portable ultrasound systems.

Proactive Resource Request for Disaster Response: A Deep Learning-Based Optimization Model

In the realm of disaster response operations, effective resource management is crucial. This research introduces an innovative approach that proactively determines the optimal quantities of resources that should be requested by local agencies. This determination is based on both current and anticipated demands, thereby ensuring a more efficient and effective response to disasters. The approach first utilizes a method that combines deep learning and temporal point process for predicting irregularly spaced future demands, and then, it formulates the resource allocation problem faced with randomly arrived demands as a stochastic optimization model. The superiority of this approach over existing resource allocation methods is demonstrated using both real-world data and simulated scenarios. The findings highlight the need for a shift from reactive to proactive strategies. Moreover, the research emphasizes the potential of advanced techniques, such as deep learning and stochastic optimization, in disaster management. These techniques can provide valuable tools for policy makers and practitioners in the field, enabling them to make more informed and effective decisions. Policies that encourage the adoption of such optimized resource allocation strategies could lead to more effective disaster response operations.

Deep learning enhanced Watershed for microstructural analysis using a boundary class semantic segmentation

The mechanical properties of the materials are determined by the size and morphology of fine microscopic features. Quantitative microstructural analysis is a key factor to establish the correlation between the mechanical properties and the thermomechanical treatment under which material condition has been achieved. As such, microstructural analysis is a very important and complex task within the manufacturing sector. Published standards are used for metallographic analysis but typically involve extensive manual interpretation of grain boundaries, resulting in measurements that are slow to produce, difficult to repeat and highly subjective. Computer vision and the evolution of artificial intelligence in the past decade can offer solutions to such problems. Deep learning and digital image processing techniques allow digital microstructural analysis to be automated using a fast and repeatable method. This paper proposes a novel boundary class semantic segmentation approach (BCSS) to identify each phase of the microstructure and additionally estimate the location of the grain boundaries. The BCSS is then combined with more traditional segmentation techniques based on the Watershed Transform to improve the identification and measurement of each feature within the microstructure using a new, hybrid automated digital microstructure analysis approach. The new method is validated on a published dataset of two-phase titanium alloy microstructure pictures captured using a scanning electron microscope. Measurements match the level of accuracy of accepted manual standards, and the method is demonstrated to be more reliable than other automated approaches. The influence of the subjective nature of manual labelling, required to train the proposed network, is also evaluated.

Go-Sharing: A Blockchain-Based Privacy-Preserving Framework for Cross-Social Network Photo Sharing

The evolution of social media has led to a trend of posting daily photos on online Social Network Platforms (SNPs). The privacy of online photos is often protected carefully by security mechanisms. However, these mechanisms will lose effectiveness when someone spreads the photos to other platforms. In this paper, we propose Go-sharing, a blockchain-based privacy-preserving framework that provides powerful dissemination control for cross-SNP photo sharing. In contrast to security mechanisms running separately in centralized servers that do not trust each other, our framework achieves consistent consensus on photo dissemination control through carefully designed smart contract-based protocols. We use these protocols to create platform-free dissemination trees for every image, providing users with complete sharing control and privacy protection. Considering the possible privacy conflicts between owners and subsequent re-posters in cross-SNP sharing, we design a dynamic privacy policy generation algorithm that maximizes the flexibility of re-posters without violating formers' privacy. Moreover, Go-sharing also provides robust photo ownership identification mechanisms to avoid illegal reprinting. It introduces a random noise black box in a two-stage separable deep learning process to improve robustness against unpredictable manipulations. Through extensive real-world simulations, the results demonstrate the capability and effectiveness of the framework across a number of performance metrics.

Artificial Intelligence and Machine Learning in Pharmacological Research: Bridging the Gap Between Data and Drug Discovery.

Artificial intelligence (AI) has transformed pharmacological research through machine learning, deep learning, and natural language processing. These advancements have greatly influenced drug discovery, development, and precision medicine. AI algorithms analyze vast biomedical data identifying potential drug targets, predicting efficacy, and optimizing lead compounds. AI has diverse applications in pharmacological research, including target identification, drug repurposing, virtual screening, de novo drug design, toxicity prediction, and personalized medicine. AI improves patient selection, trial design, and real-time data analysis in clinical trials, leading to enhanced safety and efficacy outcomes. Post-marketing surveillance utilizes AI-based systems to monitor adverse events, detect drug interactions, and support pharmacovigilance efforts. Machine learning models extract patterns from complex datasets, enabling accurate predictions and informed decision-making, thus accelerating drug discovery. Deep learning, specifically convolutional neural networks (CNN), excels in image analysis, aiding biomarker identification and optimizing drug formulation. Natural language processing facilitates the mining and analysis of scientific literature, unlocking valuable insights and information. However, the adoption of AI in pharmacological research raises ethical considerations. Ensuring data privacy and security, addressing algorithm bias and transparency, obtaining informed consent, and maintaining human oversight in decision-making are crucial ethical concerns. The responsible deployment of AI necessitates robust frameworks and regulations. The future of AI in pharmacological research is promising, with integration with emerging technologies like genomics, proteomics, and metabolomics offering the potential for personalized medicine and targeted therapies. Collaboration among academia, industry, and regulatory bodies is essential for the ethical implementation of AI in drug discovery and development. Continuous research and development in AI techniques and comprehensive training programs will empower scientists and healthcare professionals to fully exploit AI's potential, leading to improved patient outcomes and innovative pharmacological interventions.

A Literature Review of Human–AI Synergy in Decision Making: From the Perspective of Affordance Actualization Theory

The emergence of artificial-intelligence (AI)-powered information technology, such as deep learning and natural language processing, enables human to shift their behaving or working diagram from human-only to human–AI synergy, especially in the decision-making process. Since AI is multidisciplinary by nature and our understanding of human–AI synergy in decision-making is fragmented, we conducted a literature review to systematically characterize the phenomenon. Adopting the affordance actualization theory, we developed a framework to organize and understand the relationship between AI affordances, the human–AI synergy process, and the outcomes of human–AI synergy. Three themes emerged from the review: the identification of AI affordances in decision-making, human–AI synergy patterns regarding different decision tasks, and outcomes of human–AI synergy in decision-making. For each theme, we provided evidence on the existing research gaps and proposed future research directions. Our findings provide a holistic framework for understanding human–AI synergy phenomenon in decision-making. This work also offers theoretical contributions and research directions for researchers studying human–AI synergy in decision-making.

Wearable Recognition System for Complex Motions Based on Hybrid Deep‐Learning‐Enhanced Strain Sensors

Wearable recognition systems based on flexible electronics present immense potential for applications in human–machine interfaces, medical care, soft robots, etc. However, they experience challenges in terms of the nonideal consistency and stability of flexible sensors, which are responsible for detecting physical signals from human motions. These challenges hinder the improvement of recognition precision and capability in the wearable systems. Furthermore, the computational consumption for the recognition increases as more sensors are used to extensively gather information for distinguishing between complex motions. Herein, a wearable recognition system based on deep‐learning‐enhanced strain sensors for distinguishing between the complex motions of the human body is presented. A strain sensor based on peak–valley microstructures is fabricated and packaged to improve consistency and stability. Moreover, a lightweight hybrid convolutional neural network long short‐term memory model is designed to lower the computational costs of the deep learning process. In particular, by designing Butterworth filtering and Z‐score normalization algorithms, the error in feature extraction caused by sensor signal fluctuation is reduced, thereby improving the recognition accuracy of the proposed wearable system to 95.72% for seven gait motions and 100% for four different continuous series of Tai Chi forms.

RETRACTED ARTICLE: Features optimization selection in hidden layers of deep learning based on graph clustering

As it is widely known, big data can comprehensively describe the inherent laws governing various phenomena. However, the effective and efficient analysis of available data has become a major challenge in the fields of artificial intelligence, machine learning, data mining, and others. Deep learning, with its powerful learning ability and effective data-processing methods, has been extensively researched and applied in numerous academic domains. Nevertheless, the data obtained during the deep learning process often exhibits feature homogenization, resulting in highly redundant features in the hidden layers, which, in turn, affects the learning process. Therefore, this paper proposes an algorithm based on graph clustering to optimize the features of hidden layer units, with the aim of eliminating redundancy and improving learner generation.

HISP: Heterogeneous Image Signal Processor Pipeline Combining Traditional and Deep Learning Algorithms Implemented on FPGA

To tackle the challenges of edge image processing scenarios, we have developed a novel heterogeneous image signal processor (HISP) pipeline combining the advantages of traditional image signal processors and deep learning ISP (DLISP). Through a multi-dimensional image quality assessment (IQA) system integrating deep learning and traditional methods like RankIQA, BRISQUE, and SSIM, various partitioning schemes were compared to explore the highest-quality imaging heterogeneous processing scheme. The UNet-specific deep-learning processing unit (DPU) based on a field programmable gate array (FPGA) provided a 14.67× acceleration ratio for the total network and for deconvolution and max pool, the calculation latency was as low as 2.46 ms and 97.10 ms, achieving an impressive speedup ratio of 46.30× and 36.49× with only 4.04 W power consumption. The HISP consisting of a DPU and the FPGA-implemented traditional image signal processor (ISP) submodules, which scored highly in the image quality assessment system, with a single processing time of 524.93 ms and power consumption of only 8.56 W, provided a low-cost and fully replicable solution for edge image processing in extremely low illumination and high noise environments.

Strategies based on artificial intelligence for the detection of fake news

The article examines artificial intelligence (AI) strategies to combat fake news, highlighting the rise in misinformation, especially during the pandemic, and its negative impact on public decision-making. The accelerated spread of fake news in the face of truth underlines the urgency of effective detection methods. Through a systematic literature review, the use of machine learning, deep learning, and natural language processing (NLP) to automate the identification of fake news is explored, highlighting key data sets such as BuzzFeedNews, LIAR, and BS Detector, among others, essential to train detection algorithms. The study discusses various AI approaches and techniques applied to detection, including convolutional neural networks (CNN), bidirectional LSTM, and the combination of CNN with LSTM, showing significant improvements in accuracy and efficiency. However, the limitations of these techniques are pointed out, such as the volatility of the training data and the difficulty of adapting models to different misinformation contexts. The conclusion highlights AI as a vital tool against fake news, emphasizing the need to advance research and develop more sophisticated technologies to strengthen disinformation detection and protect information integrity in society. The fight against fake news is complex, but AI-based strategies show a promising path toward practical solutions.

DECIMER.ai: an open platform for automated optical chemical structure identification, segmentation and recognition in scientific publications

The number of publications describing chemical structures has increased steadily over the last decades. However, the majority of published chemical information is currently not available in machine-readable form in public databases. It remains a challenge to automate the process of information extraction in a way that requires less manual intervention - especially the mining of chemical structure depictions. As an open-source platform that leverages recent advancements in deep learning, computer vision, and natural language processing, DECIMER.ai (Deep lEarning for Chemical IMagE Recognition) strives to automatically segment, classify, and translate chemical structure depictions from the printed literature. The segmentation and classification tools are the only openly available packages of their kind, and the optical chemical structure recognition (OCSR) core application yields outstanding performance on all benchmark datasets. The source code, the trained models and the datasets developed in this work have been published under permissive licences. An instance of the DECIMER web application is available at https://decimer.ai.

A Proposed Artificial Intelligence Model for Android-Malware Detection

There are a variety of reasons why smartphones have grown so pervasive in our daily lives. While their benefits are undeniable, Android users must be vigilant against malicious apps. The goal of this study was to develop a broad framework for detecting Android malware using multiple deep learning classifiers; this framework was given the name DroidMDetection. To provide precise, dynamic, Android malware detection and clustering of different families of malware, the framework makes use of unique methodologies built based on deep learning and natural language processing (NLP) techniques. When compared to other similar works, DroidMDetection (1) uses API calls and intents in addition to the common permissions to accomplish broad malware analysis, (2) uses digests of features in which a deep auto-encoder generates to cluster the detected malware samples into malware family groups, and (3) benefits from both methods of feature extraction and selection. Numerous reference datasets were used to conduct in-depth analyses of the framework. DroidMDetection’s detection rate was high, and the created clusters were relatively consistent, no matter the evaluation parameters. DroidMDetection surpasses state-of-the-art solutions MaMaDroid, DroidMalwareDetector, MalDozer, and DroidAPIMiner across all metrics we used to measure their effectiveness.

Fully automatic deep learning-based lung parenchyma segmentation and boundary correction in thoracic CT scans.

The proposed work aims to develop an algorithm to precisely segment the lung parenchyma in thoracic CT scans. To achieve this goal, the proposed technique utilized a combination of deep learning and traditional image processing algorithms. The initial step utilized a trained convolutional neural network (CNN) to generate preliminary lung masks, followed by the proposed post-processing algorithm for lung boundary correction. First, the proposed method trained an improved 2D U-Net CNN model with Inception-ResNet-v2 as its backbone. The model was trained on 32 CT scans from two different sources: one from the VESSEL12 grand challenge and the other from AIIMS Delhi. Further, the model's performance was evaluated on a test dataset of 16 CT scans with juxta-pleural nodules obtained from AIIMS Delhi and the LUNA16 challenge. The model's performance was assessed using evaluation metrics such as average volumetric dice coefficient (DSCavg), average IoU score (IoUavg), and average F1 score (F1avg). Finally, the proposed post-processing algorithm was implemented to eliminate false positives from the model's prediction and to include juxta-pleural nodules in the final lung masks. The trained model reported a DSCavg of 0.9791 ± 0.008, IoUavg of 0.9624 ± 0.007, and F1avg of 0.9792 ± 0.004 on the test dataset. Applying the post-processing algorithm to the predicted lung masks obtained a DSCavg of 0.9713 ± 0.007, IoUavg of 0.9486 ± 0.007, and F1avg of 0.9701 ± 0.008. The post-processing algorithm successfully included juxta-pleural nodules in the final lung mask. Using a CNN model, the proposed method for lung parenchyma segmentation produced precise segmentation results. Furthermore, the post-processing algorithm addressed false positives and negatives in the model's predictions. Overall, the proposed approach demonstrated promising results for lung parenchyma segmentation. The method has the potential to be valuable in the advancement of computer-aided diagnosis (CAD) systems for automatic nodule detection.

Uncovering Semantic Inconsistencies and Deceptive Language in False News Using Deep Learning and NLP Techniques for Effective Management

In today's information age, false news and deceptive language have become pervasive, leading to significant challenges for individuals, organizations, and society as a whole. This study focuses on the application of deep learning and natural language processing (NLP) techniques to uncover semantic inconsistencies and deceptive language in false news, with the aim of facilitating effective management strategies.&#x0D; The research employs advanced deep learning models and NLP algorithms to analyze large volumes of textual data and identify patterns indicative of deceptive language and semantic inconsistencies. By leveraging the power of machine learning, the study aims to enhance the detection and classification of false news articles, enabling proactive management measures. The proposed approach not only examines the superficial aspects of false news but also delves deeper into the linguistic nuances and contextual inconsistencies that are characteristic of deceptive language. By employing advanced NLP techniques, such as sentiment analysis, topic modeling, and named entity recognition, the study strives to identify the underlying manipulative strategies employed by false news purveyors.&#x0D; The findings from this research have far-reaching implications for effective management. By accurately detecting semantic inconsistencies and deceptive language in false news, organizations can develop targeted strategies to mitigate the spread and impact of misinformation. Additionally, individuals can make informed decisions, enhancing their ability to critically evaluate news sources and protect themselves from falling victim to deceptive practices.&#x0D; In this research study, we suggest a hybrid system for detecting fake news that incorporates source analysis and machine learning techniques. Our system analyzes the language used in news articles to identify indicators of fake news and evaluates the credibility of the sources cited in the articles. We trained our system using a large dataset of news articles manually annotated as real or fake and evaluated its performance measured by common metrics like F1-score, recall, and precision. In comparison to other advanced fake news detection systems, our results show that our hybrid method has a high level of accuracy in detecting false news.