Providing machine learning (ML) over relational data is a mainstream requirement for data analytics systems. While almost all ML tools require the input data to be presented as a single table, many datasets are multi-table. This forces data scientists to join those tables first, which often leads to data redundancy and runtime waste. Recent works on "factorized" ML mitigate this issue for a few specific ML algorithms by pushing ML through joins. But their approaches require a manual rewrite of ML implementations. Such piecemeal methods create a massive development overhead when extending such ideas to other ML algorithms. In this paper, we show that it is possible to mitigate this overhead by leveraging a popular formal algebra to represent the computations of many ML algorithms: linear algebra. We introduce a new logical data type to represent normalized data and devise a framework of algebraic rewrite rules to convert a large set of linear algebra operations over denormalized data into operations over normalized data. We show how this enables us to automatically "factorize" several popular ML algorithms, thus unifying and generalizing several prior works. We prototype our framework in the popular ML environment R and an industrial R-over-RDBMS tool. Experiments with both synthetic and real normalized data show that our framework also yields significant speed-ups, up to 36x on real data.

The mammalian target of rapamycin (mTOR) is a potentially important therapeutic target in a broad range of cancer types. mTOR inhibitors such as rapamycin and its analogs (rapalogs) have been proven effective as anticancer agents in non-small cell lung cancer (NSCLC), whereas they strongly enhance phosphorylation of eukaryotic translation initiation factor 4E (eIF4E) and activation of Akt, which cause resistance to mTOR-targeted therapy after an initial response. Rapamycin induces eIF4E phosphorylation by activating MAPK-interacting kinases (Mnks), and therefore targeting Mnk/eIF4E pathway represents a potential therapeutic strategy for the treatment of NSCLC. Here, our results showed that over-expression of p-Mnk1 and p-eIF4E was significantly associated with poor overall survival of NSCLC patients and high expression of p-Mnk1 might act as an independent prognostic biomarker for these patients. Meanwhile, inhibiting Mnk1 expression by Mnk inhibitor (CGP57380) could abrogate rapalogs (RAD001)-induced eIF4E phosphorylation and Akt activation. Furthermore, combination of CGP57380 and RAD001 could induce NSCLC cells apoptosis via activating intrinsic mitochondrial pathway, and exert synergistic antitumor efficacy both in vitro and in vivo. In conclusion, combination of targeting both mTOR and Mnk/eIF4E signaling pathways to enhance effectiveness of mTOR-targeted cancer therapy might be significant innovation for the personalized treatment of NSCLC.

Data analytics using machine learning (ML) has become ubiquitous in science, business intelligence, journalism and many other domains. While a lot of work focuses on reducing the training cost, inference runtime and storage cost of ML models, little work studies how to reduce the cost of data acquisition, which potentially leads to a loss of sellers' revenue and buyers' affordability and efficiency. In this paper, we propose a model-based pricing (MBP) framework, which instead of pricing the data, directly prices ML model instances. We first formally describe the desired properties of the MBP framework, with a focus on avoiding arbitrage. Next, we show a concrete realization of the MBP framework via a noise injection approach, which provably satisfies the desired formal properties. Based on the proposed framework, we then provide algorithmic solutions on how the seller can assign prices to models under different market scenarios (such as to maximize revenue). Finally, we conduct extensive experiments, which validate that the MBP framework can provide high revenue to the seller, high affordability to the buyer, and also operate on low runtime cost.

Nuclear localization of β-catenin is essential for the progression of various human cancers via transcriptional upregulation of downstream genes. The MAP kinase interacting serine/threonine kinase (MNK)-eukaryotic translation initiation factor 4E (eIF4E) axis has been reported to activate Wnt/β-catenin signaling, and CGP57380, an inhibitor of MNK kinases, inhibits the proliferation of multiple cancers. In this study, we showed that β-catenin signaling (including β-catenin, cyclin D1, c-Myc, and MMP-7) and p-eIF4E expression were elevated in nasopharyngeal carcinoma (NPC) compared with non-cancerous nasopharyngeal epithelial tissues, and was associated with clinical characteristics of NPC patients. Lymph node metastasis, gender, aberrant β-catenin expression, and elevated levels of MMP-7 and cyclin D1 were independent prognostic factors. Significantly, expression of p-eIF4E was positively correlated with β-catenin, and targeting the MNK-eIF4E axis with CGP57380 downregulated β-catenin in the nucleus, which in turn decreased proliferation, cell cycle progression, migration, invasion, and metastasis of NPC in vitro and in vivo. CGP57380 also potentiated radiation-induced apoptosis in NPC. Moreover, CGP57380 upregulated β-catenin in the cytoplasm thus blocking epithelial-mesenchymal transition (EMT), a key mechanism in cancer cell invasiveness and metastasis. Mechanistically, inhibition of β-catenin nuclear translocation by CGP57380 was dependent on AKT activation. Notably, identification of the MNK/eIF4E/β-catenin axis might provide a potential target for overcoming the poor prognosis mediated by β-catenin in NPC.