This paper presents the framework and results from the team “Florida International University-University of Miami (FIU-UM)” in the TRECVID 2018 Ad-hoc Video Search (AVS)  task. We submitted four manually-assisted runs as follows. • M_D_FIU_UM.18_1 & M_D_FIU_UM.18_3: Convolutional Neural Network (CNN) features + linear Support Vector Machine (SVM), scores from other sources, two different sets of concepts and weighted combinations (“and”, “or”, & “mix” operations) • M_D_FIU_UM.18_2: CNN features + linear SVM, scores from other sources, weighted combination (“and”, “or”, & “mix” operations) + rectified linear score normalization • M_D_FIU_UM.18_4: CNN features + linear SVM, scores from other sources, weighted combination (“and”, “or”, & “mix” operations) + fuse different score sets (“merge” operation) Our framework includes the following processing steps: (1) manual extraction of the most important keywords based on a given query, (2) generation of CNN features from keyframes, (3) generation of scores for each concept using the linear SVM classifier, (4) generation of additional scores from multiple pre-trained models for image classification, object, scene, and action detection, (5) just-in-time concept learning for keywords not found in the concept bank, and (6) integration of the scores using the “and”, “or”, “mix”, and “merge” operators. The performance results show that our first run (M_D_FIU_UM.18_1), which includes our best-weighted combination scores, outperforms the other three runs. This year, the FIU-UM team achieved the second highest score in the manually-assisted run and ranked third among all the submitted runs (combining manually-assisted and automatic runs). The submission details are listed as follows. • Class: M (Manually-assisted runs) • Training type: D (IACC & non-IACC non-TRECVID data) • Team ID: FIU-UM (Florida International University - University of Miami) • Year: 2018.