Retrosynthesis is the process of identifying a set of reactants to synthesize a target molecule. It is critical to material design and drug discovery. Existing machine learning approaches based on language models and graph neural networks have achie rarely ved discussed, encouraging and rigorous results. Ho evaluations wever, the of inner these connections models are of lar these gely in models need.

Conformal selection (CS) uses calibration data to identify test inputs whose unobserved outcomes are likely to satisfy a pre-specified minimal quality requirement, while controlling the false discovery rate (FDR). Existing methods fix the target FDR level before observing data, which prevents the user from adapting the balance between number of selected test inputs and FDR to downstream needs and constraints based on the available data. For example, in genomics or neuroimaging, researchers often inspect the distribution of test statistics, and decide how aggressively to pursue candidates based on observed evidence strength and available follow-up resources. To address this limitation, we introduce {post-hoc CS} (PH-CS), which generates a path of candidate selection sets, each paired with a data-driven false discovery proportion (FDP) estimate. PH-CS lets the user select any operating point on this path by maximizing a user-specified utility, arbitrarily balancing selection size and FDR. Building on conformal e-variables and the e-Benjamini-Hochberg (e-BH) procedure, PH-CS is proved to provide a finite-sample post-hoc reliability guarantee whereby the ratio between estimated FDP level and true FDP is, on average, upper bounded by $1$, so that the average estimated FDP is, to first order, a valid upper bound on the true FDR. PH-CS is extended to control quality defined in terms of a general risk. Experiments on synthetic and real-world datasets demonstrate that, unlike CS, PH-CS can consistently satisfy user-imposed utility constraints while producing reliable FDP estimates and maintaining competitive FDR control.

In this way, even with limited meta samples, MTG-Net holds the potential to produce reasonable gain estimations on arbitrary task combinations.

Experimental results showthattheproposed metriclearning algorithm improves both certified robust errors and empirical robust errors (errors under adversarial attacks).

We fully understand the concern about our baselines since2 we are the first to improve certified robustness of metric learning. Therefore, as Reviewer 4 suggested, we add3 experiments comparing with neural networks certification methods, including ordinary neural networks certified by4 CROWN [48] and randomized-smoothing neural networks [11]. The results are shown in Figure i. In general, computational11 cost is not an issue for ARML. To make the comparison fair, all of15 the methods are run on CPU (Xeon(R) E5-2620 v4 @16 2.10GHz).

ROUGE-N scores are the overlapping of n-grams [2] between the generated and referencesequence. Those scores are then averaged overthe whole corpus toreach anoverall quality. For both proposed MMMMI attack methods, shadow models are indispensable. The first hidden layer in the attack model has 256 units and the second hidden layer has20units, bothactivatedbyReLU function. We used resnet-LSTM architecture as the target model architecture.