Combinatorial CRISPR screening enables large-scale identification of synergistic gene pairs for combination therapies, but exhaustive experimentation is infeasible. We introduce NAIAD, an active learning framework that efficiently discovers optimal gene pairs by leveraging single-gene perturbation effects and adaptive gene embeddings that scale with the training data size, mitigating overfitting in small-sample learning while capturing complex gene interactions as more data is collected. Evaluated on four CRISPR datasets with over 350,000 interactions, NAIAD trained on small datasets outperforms existing models by up to 40%. Its recommendation system prioritizes gene pairs with maximum predicted effects, accelerating discovery with fewer experiments. We also extend NAIAD to optimal drug combination identification among 2,000 candidates. Overall, NAIAD enhances combinatorial perturbation design and drives advances in genomics research and therapeutic development in combination therapy. Our code is publicly available at: https://github.com/NeptuneBio/NAIAD

Humans exhibit a remarkable capacity for anticipating the actions of others and planning their own actions accordingly. In this study, we strive to replicate this ability by addressing the social motion prediction problem. We introduce a new benchmark, a novel formulation, and a cognition-inspired framework. We present Wusi, a 3D multi-person motion dataset under the context of team sports, which features intense and strategic human interactions and diverse pose distributions. By reformulating the problem from a multi-agent reinforcement learning perspective, we incorporate behavioral cloning and generative adversarial imitation learning to boost learning efficiency and generalization. Furthermore, we take into account the cognitive aspects of the human social action planning process and develop a cognitive hierarchy framework to predict strategic human social interactions. We conduct comprehensive experiments to validate the effectiveness of our proposed dataset and approach.

Early detection of cancer is crucial to treatment. In this paper, we investigated an image based fractal analysis method for detecting cancer cells. Cancer cells typically display abnormalities including unregulated cell growth. Fractal analysis can be used to analyze irregularly shaped figures and measure morphological complexity. Experiments with images of human breast cancer cells were investigated. Changes in the fractal dimension between cancer cells and healthy cells were analyzed and compared. Our preliminary results show that the image based fractal analysis technique is able to detect breast cancer cells. It has great potential for yielding insight into the morphological characterization of tumor growth and may serve as an indicator for cancer detection and cancer treatment efficiency.

By analyzing a set of access attempts by teenagers to pornographic Web sites, we found that more than half of them are image searches and visits to Web sites with little text information. It is obvious that textual content-based filters cannot correctly categorize such access attempts. This paper describes a novel URL-based objectionable content categorization approach and its application to Web filtering. In this approach, we break the URL into a sequence of n-grams with a range of n's and then a machine learning algorithm is applied to the n-gram representation of URLs to learn a classifier of pornographic Web sites. We showed empirically that the URL-based approach is able to correctly identify many of the objectionable Web pages. We also demonstrated that the optimum Web filtering results could be achieved when it was used with a content-based approach in a production environment