The saliency methods are widely used for generating heatmaps that emphasize the important portions of an input image for deep networks on a specific classification task. Interpretability is crucial for deploying deep ...
The saliency methods are widely used for generating heatmaps that emphasize the important portions of an input image for deep networks on a specific classification task. Interpretability is crucial for deploying deep neural networks in real-world applications. However, the heatmaps produced by current visual explainable methods may contain or visualize particulars differently. To analyze and compare the visualization of different methods, such as Gradient-based, Activation-based, Perturbation-based, and Region-based methods, we empirically evaluated them on the acute lymphoblastic leukemia (cancer cell) classification task using state-of-the-art convolutional neural networks. We also visualized the essential pathological features (salient parts) that are the reasons for the classification results on the classification of normal versus malignant cell (CNMC) dataset.
In the creation of Hopf topological matters, the old paradigm is to conceive the Hopf invariant first, and then display its intuitive topology through links. Here we brush aside this effort and put forward a recipe fo...
详细信息
In the creation of Hopf topological matters, the old paradigm is to conceive the Hopf invariant first, and then display its intuitive topology through links. Here we brush aside this effort and put forward a recipe for unraveling the quenched two-dimensional (2D) two-band Chern insulators under a parallel quench protocol, which implies that the quench quantities with different momentum k are parallel or antiparallel to each other. We find that whether the dynamical Hopf invariant exists or not, the links in (2+1)D space always keep their standard shape even for topological initial states, and trace out the trajectories of phase vortices. The linking number is exactly equal to the difference between pre- and postquench Chern numbers regardless of the construction of homotopy groups. We employ two concrete examples to illustrate these results, highlighting the polarity reversal at fixed points.
Individual preferences change over time, requiring recommendation systems that adapt and provide personalized suggestions. This paper introduces a novel approach called Preference Tracing, inspired by knowledge tracin...
详细信息
ISBN:
(数字)9798350386097
ISBN:
(纸本)9798350386103
Individual preferences change over time, requiring recommendation systems that adapt and provide personalized suggestions. This paper introduces a novel approach called Preference Tracing, inspired by knowledge tracing from the educational domain. Knowledge tracing estimates a student’s knowledge state from interactions with question-response pairs and knowledge components, which are essential for solving given exercises. Based on the estimated knowledge state, the model predicts the probability of correctly answering subsequent exercises. Similarly, Preference Tracing estimates a user’s preference state from rating histories, including movie-rating pairs and a movie component. Movie plots were crawled from Wikipedia, IMDb, and Letterboxd, and then latent Dirichlet allocation (LDA) was applied to define each film’ s top-weighted topic as a movie component. Based on that, Preference Tracing can track users’ changing preferences and predict whether a user would like a given movie. Our main contribution demonstrates that Preference Tracing delivers hyper-personalized recommendations by adapting to changing individual preferences. Experimental results on MovieLens 1M show that Preference Tracing outperforms traditional baseline models and effectively captures dynamic changes.
Intermediate Representations (IRs) are essential in compiler design and program analysis, yet their comprehension by Large Language Models (LLMs) remains underexplored. This paper presents a pioneering empirical study...
详细信息
In this paper, a novel broadband high-selectivity stacked filtering antenna with multiple radiation nulls is proposed. The antenna consists of a short-circuited stepped impedance feeding line and two U-shaped strips o...
详细信息
We propose a novel AR CAPTCHA that first uses Augmented Reality to design CAPTCHA. Users should use their cameras on mobile devices to capture a marker in the 3D physical world or PC screens to find the appropriate an...
详细信息
This paper evaluates the multiple-input multiple-output underlay cognitive multihop relay networks with short-packet communications, where general and practical scenarios are considered with multiple primary users and...
详细信息
In this paper, we discuss a new p-Laplacian fractional differential equation involving instantaneous and non-instantaneous impulses, supplemented with Sturm-Liouville boundary conditions. To study the stated problem, ...
详细信息
OBJECTIVE To establish a prediction model of coronary heart disease(CHD)in elderly patients with diabetes mellitus(DM)based on machine learning(ML)*** Based on the Medical bigdata Research Centre of Chinese PLA Gener...
详细信息
OBJECTIVE To establish a prediction model of coronary heart disease(CHD)in elderly patients with diabetes mellitus(DM)based on machine learning(ML)*** Based on the Medical bigdata Research Centre of Chinese PLA General Hospital in Beijing,China,we identified a cohort of elderly inpatients(≥60 years),including 10,533 patients with DM complicated with CHD and 12,634 patients with DM without CHD,from January 2008 to December *** collected demographic characteristics and clinical *** selecting the important features,we established five ML models,including extreme gradient boosting(XGBoost),random forest(RF),decision tree(DT),adaptive boosting(Adaboost)and logistic regression(LR).We compared the receiver operating characteristic curves,area under the curve(AUC)and other relevant parameters of different models and determined the optimal classification *** model was then applied to 7447 elderly patients with DM admitted from January 2018 to December 2019 to further validate the performance of the *** Fifteen features were selected and included in the ML *** classification precision in the test set of the XGBoost,RF,DT,Adaboost and LR models was 0.778,0.789,0.753,0.750 and 0.689,respectively;and the AUCs of the subjects were 0.851,0.845,0.823,0.833 and 0.731,*** the XGBoost model with optimal performance to a newly recruited dataset for validation,the diagnostic sensitivity,specificity,precision,and AUC were 0.792,0.808,0.748 and 0.880,*** The XGBoost model established in the present study had certain predictive value for elderly patients with DM complicated with CHD.
Forested areas are extremely vulnerable to disasters leading to environmental *** Fire is one among them which requires immediate *** are lot of works done by authors where Wireless Sensors and IoT have been used for ...
详细信息
Forested areas are extremely vulnerable to disasters leading to environmental *** Fire is one among them which requires immediate *** are lot of works done by authors where Wireless Sensors and IoT have been used for forest fire ***,towards monitoring the forest fire and managing the energy efficiently in IoT,Energy Efficient Routing Protocol for Low power lossy networks(E-RPL)was *** were challenges about the scalability of the network resulting in a large end-to-end delay and less packet delivery which led to the development of Aggregator-based Energy Efficient RPL with data Compression(CAAERPL).Though CAA-ERPL proved effective in terms of reduced packet delivery,less energy consumption,and increased packet delivery ratio for varying number of nodes,there is still challenge in the selection of aggregator which is based purely on probability percentage of *** has been research work where fuzzy logic been employed for Mobile Ad-hoc Routing,RPL routing and cluster head selection in Wireless *** has been no work where fuzzy logic is employed for aggregator selection in Energy Efficient *** accordingly,we here have proposed Fuzzy Based Aggregator selection in Energy-efficient RPL for region thereby forming DODAG for communicating to Fog/*** here have developed fuzzy inference rules for selecting the aggregator based on strength which takes residual power,Node degree,and Expected Transmission Count(ETX)as input *** Fuzzy Aggregator Energy Efficient RPL(FA-ERPL)based on fuzzy inference rules were analysed against E-RPL in terms of scalability(First and Half Node die),Energy Consumption,and aggregator node energy *** the analysis,it was found that FA-ERPL performed better than *** were simulated using MATLAB and results.
暂无评论