Click-through rate (CTR) prediction is a critical task in online display advertising. The data involved in CTR prediction are typically multi-field categorical data, i.e., every feature is categorical and belongs to one and only one field. One of the interesting characteristics of such data is that features from one field often interact differently with features from different other fields. Recently, Fieldaware Factorization Machines (FFMs) have been among the best performing models for CTR prediction by explicitly modeling such difference. However, the number of parameters in FFMs is in the order of feature number times field number, which is unacceptable in the real-world production systems. In this paper, we propose Field-weighted Factorization Machines (FwFMs) to model the different feature interactions between different fields in a much more memory-efficient way. Our experimental evaluations show that FwFMs can achieve competitive prediction performance with only as few as 4% parameters of FFMs. When using the same number of parameters, FwFMs can bring 0.92% and 0.47% AUC lift over FFMs on two real CTR prediction data sets.

The stabilization of liquid-metal foams by ceramic particles is studied by experimental simulations. The objective is to determine the optimum wetting property for liquid-foam stability. Ceramic particles are mimicked by inert plastic particles. The liquid metal is mimicked by a continuous, surfactantfree ethanol-water solution. The wetting property of the plastic particles in the liquid solution is changed continuously by varying the liquid composition. The experimental simulation shows that the liquid-foam stabilization by the solid particles depends strongly on the wetting property. An optimum wetting-angle range of 75 to 85 deg is determined from the experiments. The foam stability is shown to be unrelated to liquid viscosity, which remains unchanged with the wetting angle. Foams formed in the optimum wetting condition exhibit a slow decay, a stable foam volume that persists for a long time, and a fine cell structure in the micrometer range. The selection of ceramic particles for optimal stabilization of liquid-metal foams and the foam-processing procedures are discussed in the light of these experimental simulation results.

Conversion prediction plays an important role in online advertising since Cost-Per-Action (CPA) has become one of the primary campaign performance objectives in the industry. Unlike click prediction, conversions have different types in nature, and each type may be associated with different decisive factors. In this paper, we formulate conversion prediction as a multi-task learning problem, so that the prediction models for different types of conversions can be learned together. These models share feature representations, but have their specific parameters, providing the benefit of information-sharing across all tasks. We then propose Multi-Task Field-weighted Factorization Machine (MT-FwFM) to solve these tasks jointly. Our experiment results show that, compared with two state-of-the-art models, MT-FwFM improve the AUC by 0.74% and 0.84% on two conversion types, and the weighted AUC across all conversion types is also improved by 0.50%.

The frog, Xenopus laevis, possesses a high capacity to regenerate various larval tissues, including the lens, which is capable of complete regeneration from the cornea epithelium. However, the molecular signaling mechanisms of cornea-lens regeneration are not fully understood. Previous work has implicated the involvement of the Wnt signaling pathway, but molecular studies have been very limited. Iris-derived lens regeneration in the newt (Wolffian lens regeneration) has shown a necessity for active Wnt signaling in order to regenerate a new lens. Here we provide evidence that the Wnt signaling pathway plays a different role in the context of cornea-lens regeneration in Xenopus. We examined the expression of frizzled receptors and wnt ligands in the frog cornea epithelium. Numerous frizzled receptors (fzd1, fzd2, fzd3, fzd4, fzd6, fzd7, fzd8, and fzd10) and wnt ligands (wnt2b.a, wnt3a, wnt4, wnt5a, wnt5b, wnt6, wnt7b, wnt10a, wnt11, and wnt11b) are expressed in the cornea epithelium, demonstrating that this tissue is transcribing many of the ligands and receptors of the Wnt signaling pathway. When compared to flank epithelium, which is lens regeneration incompetent, only wnt11 and wnt11b are different (present only in the cornea epithelium), identifying them as potential regulators of cornea-lens regeneration. To detect changes in canonical Wnt/β-catenin signaling occurring within the cornea epithelium, axin2 expression was measured over the course of regeneration. axin2 is a well-established reporter of active Wnt/β-catenin signaling, and its expression shows a significant decrease at 24 hours post-lentectomy. This decrease recovers to normal endogenous levels by 48 hours. To test whether this signaling decrease was necessary for lens regeneration to occur, regenerating eyes were treated with either 6-bromoindirubin-3’-oxime (BIO) or 1-azakenpaullone – both activators of Wnt signaling – resulting in a significant reduction in the percentage of cases with successful regeneration. In contrast, inhibition of Wnt signaling using either the small molecule IWR-1, treatment with recombinant human Dickkopf-1 (rhDKK1) protein, or transgenic expression of Xenopus DKK1, did not significantly affect the percentage of successful regeneration. Together, these results suggest a model where Wnt/β-catenin signaling is active in the cornea epithelium and needs to be suppressed during early lens regeneration in order for these cornea cells to give rise to a new lentoid. While this finding differs from what has been described in the newt, it closely resembles the role of Wnt signaling during the initial formation of the lens placode from the surface ectoderm during early embryogenesis.