第 2 / 3 页
非代码面试题
显示 20 / 41 道匹配题目
答题状态：未尝试未正确已正确
ID题目领域难度题型进度权限
2510Zero Lambda Recovers OLS 16Why do ridge and lasso both reduce to OLS when their regularization parameter is set to zero?机器学习困难derivation未尝试面试订阅2521Intercept-Only Logistic MLE 1For an intercept-only logistic model with n 1 positives and n 0 negatives, what fitted probability p hat maximizes the log-likelihood?机器学习简单derivation未尝试免费2522Intercept From the Positive Rate 2In an intercept-only logistic model, if the fitted probability is p hat, what intercept b solves sigma(b)=p hat?机器学习简单derivation未尝试免费2523Gradient of Logistic Negative Log-Likelihood 3For one observation (x,y) with y in 0,1 and score z = w T x, what is the gradient of the negative log-likelihood with respect to w?机器学习中等derivation未尝试免费2524Why No Closed Form in Logistic Regression 5Why does logistic regression usually require iterative optimization rather than a normal-equation-style closed form?机器学习中等essay未尝试免费2526Why Separable Data Pushes Coefficients Outward 7Why do logistic-regression coefficients tend to diverge on perfectly linearly separable data if no regularization is used?机器学习简单essay未尝试免费2528Why Log-Loss Rewards Calibration 9Why does a well-calibrated probability forecaster typically fare better under log-loss than a forecaster that only gets rankings right?机器学习中等essay未尝试免费2529Why Regularization Helps Even When Logistic Is Convex 11If logistic loss is already convex, why can regularization still be crucial in practice?机器学习中等essay未尝试面试订阅2532Hessian of Logistic Negative Log-Likelihood 4For one observation with score z = w T x, what is the Hessian of the negative log-likelihood with respect to w?机器学习简单derivation未尝试免费2537Why Logistic Probabilities Are Useful Downstream 18Why is it valuable that logistic regression produces a calibrated probability estimate rather than only a hard class label?机器学习中等essay未尝试面试订阅2538Why Logistic Beats Hard Threshold Rules for Training 23Why is a smooth probabilistic loss easier to optimize than training directly against a hard classification rule?机器学习中等essay未尝试免费5016Infer Competing Density From Posterior Responsibility 1In a two-component Gaussian mixture, component 1 has weight 0.6, component 2 has weight 0.4, and the density of component 1 at an observation x is g1(x)=0.3. The E-step assigns posterior responsibility 0.75 to component 1. What density value g2(x) for component 2 is implied?机器学习简单数值题未尝试面试订阅5021Recover Missing Observation From EM Mean Update 6A Gaussian-mixture M-step updates one component mean to 0.875. Three observations contribute to that component with responsibilities 0.8, 0.6, and 0.1. Two observed values are 0 and 1, while the third observation x is missing from the log. What missing x is implied by the updated mean?机器学习中等数值题未尝试面试订阅5023Recover Missing Observation From EM Mean Update 8A Gaussian-mixture M-step updates one component mean to 0.4. Three observations contribute to that component with responsibilities 0.4, 0.4, and 0.2. Two observed values are -1 and 1, while the third observation x is missing from the log. What missing x is implied by the updated mean?机器学习中等数值题未尝试面试订阅5026Recover Effective Membership From Updated Mixture Weight 11An EM fit is run on 80 observations, and after the M-step the updated weight of one component is 0.35. What total responsibility mass was assigned to that component?机器学习简单数值题未尝试面试订阅5031BIC Improvement Threshold for an Extra Component 16A mixture model with one extra component would add Delta k = 3 free parameters on a sample of size n = 100. Under BIC = -2 log L + k log n, what minimum log-likelihood improvement Delta log L is needed before the larger model is preferred?机器学习中等数值题未尝试面试订阅5036EM Mixture Diagnostic 21Why does one EM component sometimes collapse onto a single point with near-zero variance during mixture training?机器学习困难essay未尝试面试订阅5037EM Mixture Diagnostic 22Why can two EM runs on the same data land on noticeably different mixture parameters even when they reach similar likelihoods?机器学习困难essay未尝试面试订阅5038EM Mixture Diagnostic 23Why is label switching not a bug in a mixture model but still a headache when you compare runs?机器学习困难essay未尝试面试订阅5039EM Mixture Diagnostic 24Why can k-means and a Gaussian mixture with shared spherical covariances give similar clusters, yet still disagree on borderline points?机器学习困难essay未尝试面试订阅