Recent advances in biosensing technologies have led to applications of biosensor probe arrays for rapid identification of biological agents such as drugs, gene expressions, proteins, cholesterol and fats in an input sample. However, monitoring the simultaneous presence of multiple agents in a sample is still a challenging task. Multiple agents may often attach to the same probes, leading to low specificity. By using microarrays as a specific example, we introduce two methods based on conditional deduction and non-unique probes to detect multiple targets. We introduce three quality metrics, namely: effectiveness, cost and reliability to evaluate different designs of microarrays and propose two ILP/Pseudo-Boolean models for optimizing on these metrics. By applying on various synthetic and real datasets, we demonstrate the importance of these quality metrics in designing microarrays for multiple target detections.