Estimating the effect of an intervention (i.e. causal inference) is a fundamental problem in fields like medicine, e-commerce, government (among others). An ideal approach requires experimental data, which is often ha...
详细信息
ISBN:
(数字)9781728162515
ISBN:
(纸本)9781728162522
Estimating the effect of an intervention (i.e. causal inference) is a fundamental problem in fields like medicine, e-commerce, government (among others). An ideal approach requires experimental data, which is often hard to find or collect. Observational data on the other hand is in abundance, but contains systematic bias posing a great challenge for this task. In this paper, we introduce the use of two methods pioneered in domain adaptation: Euclidean and non-Euclidean correlation alignment in the context of causal inference. We evaluate their performance using widely used datasets and our results perform favourably against the current state of the art.
The paper compares and discusses the impact of distributed deep learning training using the Horovod framework on the performance metrics of image segmentation model trained using the Mask R-CNN (Region-based Convoluti...
详细信息
In this white paper, we synthesize key points made during presentations and discussions from the AI-Assisted Decision Making for Conservation workshop, hosted by the Center for Research on Computation and Society at H...
详细信息
Brain tumor segmentation plays a crucial role in medical image analysis. Brain tumor patients considerably benefit from early discovery due to the increased likelihood of a successful outcome from therapy. Due to the ...
Brain tumor segmentation plays a crucial role in medical image analysis. Brain tumor patients considerably benefit from early discovery due to the increased likelihood of a successful outcome from therapy. Due to the sheer volume of MRI images generated in everyday clinical practice, manually isolating brain tumors for cancer diagnosis is a challenging task. Automatic segmentation of images of brain tumors is essential. This system aimed to synthesize previous methods for BSA-RELM-based brain tumor segmentation. The proposed methodology rests on four fundamental pillars: preprocessing, segmentation, feature extraction, and model training. Filtering, scaling, boosting contrast, and sharpening are all examples of preprocessing techniques. When doing segmentation, a clustering technique based on Fuzzy Clustering Means (FCM) is used to breakdown the overall dataset into numerous subsets. The proposed approach used the region of filling for feature extraction. After that, a BSA-RELM is used to train the models with the input features. The proposed technique outperforms BSA and RELM, two of the most common alternatives. There was a 98.61 percent success rate with the recommended method.
The occurrence of an adverse drug event (ADE) has become a serious social concern of public health. Early detection of ADEs can lower the risk of drug safety as well as the expense of the drug. While post-market spont...
This paper examines the mixed convective heat transfer (HTR) of nanofluid (NFD) flow in a rectangular enclosure with the upper moving wall numerically. The lower wall has a high temperature and a number of semi-circul...
详细信息
This paper examines the mixed convective heat transfer (HTR) of nanofluid (NFD) flow in a rectangular enclosure with the upper moving wall numerically. The lower wall has a high temperature and a number of semi-circular obstacles with the same temperature are installed on it. The upper moving wall has a low temperature and the other two walls are insulated. The enclosure can change from horizontal to vertical. Radiation HTR is considered in the enclosure and there is a magnetic field (MGF) that can change the angle from horizontal to vertical affecting the NFD. This study is carried out for different angles of the enclosure and MGF from horizontal to vertical for radiation parameters (RDP) of 0 to 3 and a constant MGF with Hartmann number of 20 and Richardson number of 10. The aim is to estimate the Nusselt number (Nu), entropy generation (ETG), and Bejan number (Be). The SIMPLE algorithm is utilized using FORTRAN software, and optimization is done using artificial intelligence to find the maximum and minimum output values. The results demonstrate that the maximum value of Nu and Bes corresponds to the MGF angle and enclosure angle of 90°. The minimum value of the Nu and the maximum amount of ETG corresponds to the horizontal MGF and horizontal enclosure when the RDP is 1.5. An increment in the RDP enhances the amount of Nu. The maximum amount of ETG, i.e. 12.87, corresponds to the enclosure with an angle of 45° for the horizontal MGF and the absence of RDP. corresponds to the enclosure with an angle of 45° for the horizontal MGF and the absence of RDP. It was also found that most environmental impacts, and hence values for different environmental factors, arise from the production of nanoparticles; thus, it is a significant contributor to environmental impacts.
Cancer that originates in the breast tissue then spreads to the chest wall is called breast cancer. Doctors routinely examine mammograms for signs of cancer; however, aberrant macrocalcifications and microcalcificatio...
详细信息
Cancer that originates in the breast tissue then spreads to the chest wall is called breast cancer. Doctors routinely examine mammograms for signs of cancer; however, aberrant macrocalcifications and microcalcifications might appear on mammograms when the picture quality is subpar. Always get checked out if you see anything out of the ordinary, especially if it involves your breasts, such abnormal calcium deposits. For this mammographic deposit to be properly interpreted, top-notch picture quality is necessary. Many different breast cancer screening methods and the many breast cancer phases are still the subject of active study. In order to construct effective medical image processing systems, experts use methods including the Ant Colony Algorithm (ACA), the Improved Adaptive Fuzzy C-Means (IAFCM), and TNM (the size of the breast tumor (T), the lymph nodes around the tumor, and metastasized). Classes were determined using an MPIG, or a modified Poisson inverse gradient classifier. More than five hundred picture modalities are used across all methods. Medical professionals that rely on images to establish diagnoses or treatments might find the results of this research useful.
In response to the formidable challenges surrounding data security, particularly in the face of evolving threats and vulnerabilities, this study introduces the Advanced Blockchain-Enhanced data Security Protocol (ABE-...
详细信息
ISBN:
(数字)9798350395327
ISBN:
(纸本)9798350395334
In response to the formidable challenges surrounding data security, particularly in the face of evolving threats and vulnerabilities, this study introduces the Advanced Blockchain-Enhanced data Security Protocol (ABE-DSP) as a cutting-edge solution. Acknowledging the complexities of safeguarding sensitive information, ABE-DSP is a meticulously crafted and sophisticated framework that aims to fortify the security and integrity of data transactions through the integration of blockchain technology. This protocol addresses the pressing challenge of ensuring robust data security by placing a primary focus on enhancing overall system security and refining user authentication processes. Within the context of these challenges, the study explores the intricacies of ABE-DSP's design and implementation. Furthermore, the proposed method demonstrates its effectiveness by achieving a remarkable success rate, processing 97% of received packets with precision. This high percentage underscores the protocol's efficacy and reliability, positioning it as a robust and advanced tool for overcoming the contemporary challenges associated with data security. This study contributes valuable insights to the ongoing discourse on leveraging blockchain technology for identity-centric data security, offering a promising avenue for addressing and mitigating the challenges faced in the dynamic landscape of information protection.
While state-of-the-art NLP models have been achieving the excellent performance of a wide range of tasks in recent years, important questions are being raised about their robustness and their underlying sensitivity to...
详细信息
Multi-turn dialogue generation aims to generate natural and fluent responses that should be consistent with multiple consecutive utterances history. It is a more challenging task compared to its single-turn counterpar...
详细信息
ISBN:
(纸本)9781665438599
Multi-turn dialogue generation aims to generate natural and fluent responses that should be consistent with multiple consecutive utterances history. It is a more challenging task compared to its single-turn counterpart since it requires the model to capture the topic drift along with the multi-turn dialogue history. In this paper, we propose a multi-turn dialogue generation model which incorporates topic drift aware information into a hierarchical encoder-decoder framework to generate coherent responses. This model first utilizes a Convolutional Neural Network (CNN) based topic model to obtain the topic representation of each utterance. Then a topic drift model is employed to encode the sequential topics of multi-turn dialogue history to infer the topic of response. During the response generation, a specially designed topic drift aware generator is proposed to dynamically balance the impact of the inferred topic of response and local word structure. Fur-thermore, we employ multi-task learning to optimize the topic drift model and dialogue generation simultaneously. Extensive experimental results on two benchmark datasets (i.e. Cornell Movie Dialog Corpus and Ubuntu Dialogue dataset) indicate that our proposed model can generate more coherent responses, and significantly outperform other dialogue generation models.
暂无评论