Flipped classroom teaching puts forward new requirements on the enthusiasm of students’ independent learning, however, the traditional independent learning lacks scientific aids and cannot meet the individual needs of students in the process of self-study. Therefore, this paper exploits the neural network technology in intelligent computing technology to extract the deep implicit semantic representation, combines the implicit semantic indexing (LSI) to improve the traditional collaborative filtering algorithm, and explores an optimized implementation path of the flipped classroom teaching mode. The improved ICF algorithm outperforms the comparison algorithm in terms of recommendation accuracy, average recall, and average coverage in the three datasets. The computational time consumed is reduced by 44.85%, 57.34%, and 73.68%, respectively, compared with UCF. Incorporating the learning resource recommendation model constructed in this paper in a traditional flipped classroom, it is found that the post-test scores of the experimental class in Moral Education are significantly higher than those of the control class (p<0.01), and its post-test scores are significantly higher than its pre-test scores (p<0.01). The collaborative filtering algorithm optimized by intelligent computing technology facilitates students' personalized independent learning, innovates the general flipped classroom teaching mode, and receives the expected results.