Search

We introduce STREET, a unified multi-task and multi-domain natural language reasoning and explanation benchmark. Unlike most existing question-answering (QA) datasets, we expect models to not only answer questions, but also produce step-by-step structured explanations describing how premises in the question are used to produce intermediate conclusions that can prove the correctness of a certain answer. We perform extensive evaluation with popular language models such as few-shot prompting GPT-3 and fine-tuned T5. We find that these models still lag behind human performance when producing such structured reasoning steps. We believe this work will provide a way for the community to better train and test systems on multi-step reasoning and explanations in natural language., Comment: Published in ICLR 2023

In online advertising, it is critical for advertisers to forecast conversion rate (CVR) of campaigns. Previous work on campaign forecasting concentrates on the time-series analysis which depend on the availability of a length of history. However, these approaches become inadequate for cold-start campaigns which lack for the observation of past. In this work, we attempt to mitigate this challenge by learning an unsupervised and composite campaign embedding to capture multi-view semantic relationships on campaign information, and consequently forecasting the cold-start campaigns using the nearest neighbor campaigns. Specifically, we propose a novel embedding framework which simultaneously extracts and fuses heterogeneous knowledge from multiple views of campaign data in a multi-task learning fashion, to learn the semantic relationship of ad message, conversion rule, and audience targeting. We develop a hierarchical attention mechanism to refine the embedding model at two levels - an intra-view attention to improve context aggregation, and an inter-task attention to balance task importance. Finally, we adopt the k-NN regression model to predict the CVR based on the neighboring campaigns in the embedding space which encodes the multi-view campaign proximity. We conduct extensive experiments on a real-world advertising campaign dataset. The results demonstrate the effectiveness of the proposed embedding method for CVR forecasting in cold-start scenarios. © 2022 IEEE.

This paper presents a novel teachable conversation interaction system that is capable of learning users preferences from cold start by gradually adapting to personal preferences. In particular, the TAI system is able to automatically identify and label user preference in live interactions, manage dialogue flows for interactive teaching sessions, and reuse learned preference for preference elicitation. We develop the TAI system by leveraging BERT encoder models to encode both dialogue and relevant context information, and build action prediction (AP), argument filling (AF) and named entity recognition (NER) models to understand the teaching session. We adopt a seeker-provider interaction loop mechanism to generate diverse dialogues from cold-start. TAI is capable of learning user preference, which achieves 0.9122 turn level accuracy on out-of-sample dataset, and has been successfully adopted in production., Comment: preference, personalization, cold-start, dialogue, LLM. embedding

Existing neural relevance models do not give enough consideration for query and item context information which diversifies the search results to adapt for personal preference. To bridge this gap, this paper presents a neural learning framework to personalize document ranking results by leveraging the signals to capture how the document fits into users' context. In particular, it models the relationships between document content and user query context using both lexical representations and semantic embeddings such that the user's intent can be better understood by data enrichment of personalized query context information. Extensive experiments performed on the search dataset, demonstrate the effectiveness of the proposed method., Comment: Contextual, Personalization, Search, Semantics, LLM, embedding

Consensus clustering aggregates partitions in order to find a better fit by reconciling clustering results from different sources/executions. In practice, there exist noise and outliers in clustering task, which, however, may significantly degrade the performance. To address this issue, we propose a novel algorithm -- robust consensus clustering that can find common ground truth among experts' opinions, which tends to be minimally affected by the bias caused by the outliers. In particular, we formalize the robust consensus clustering problem as a constraint optimization problem, and then derive an effective algorithm upon alternating direction method of multipliers (ADMM) with rigorous convergence guarantee. Our method outperforms the baselines on benchmarks. We apply the proposed method to the real-world advertising campaign segmentation and forecasting tasks using the proposed consensus clustering results based on the similarity computed via Kolmogorov-Smirnov Statistics. The accurate clustering result is helpful for building the advertiser profiles so as to perform the forecasting., Comment: 8 pages

In cost-per-click (CPC) or cost-per-impression (CPM) advertising campaigns, advertisers always run the risk of spending the budget without getting enough conversions. Moreover, the bidding on advertising inventory has few connections with propensity one that can reach to target cost-per-acquisition (tCPA) goals. To address this problem, this paper presents a bid optimization scenario to achieve the desired tCPA goals for advertisers. In particular, we build the optimization engine to make a decision by solving the rigorously formalized constrained optimization problem, which leverages the bid landscape model learned from rich historical auction data using non-parametric learning. The proposed model can naturally recommend the bid that meets the advertisers' expectations by making inference over advertisers' historical auction behaviors, which essentially deals with the data challenges commonly faced by bid landscape modeling: incomplete logs in auctions, and uncertainty due to the variation and fluctuations in advertising bidding behaviors. The bid optimization model outperforms the baseline methods on real-world campaigns, and has been applied into a wide range of scenarios for performance improvement and revenue liftup.

In computational advertising, a challenging problem is how to recommend the bid for advertisers to achieve the best return on investment (ROI) given budget constraint. This paper presents a bid recommendation scenario that discovers the concavity changes in click prediction curves. The recommended bid is derived based on the turning point from significant increase (i.e. concave downward) to slow increase (convex upward). Parametric learning based method is applied by solving the corresponding constraint optimization problem. Empirical studies on real-world advertising scenarios clearly demonstrate the performance gains for business metrics (including revenue increase, click increase and advertiser ROI increase)., Comment: 10 pages

Multilingual information retrieval (IR) is challenging since annotated training data is costly to obtain in many languages. We present an effective method to train multilingual IR systems when only English IR training data and some parallel corpora between English and other languages are available. We leverage parallel and non-parallel corpora to improve the pretrained multilingual language models' cross-lingual transfer ability. We design a semantic contrastive loss to align representations of parallel sentences that share the same semantics in different languages, and a new language contrastive loss to leverage parallel sentence pairs to remove language-specific information in sentence representations from non-parallel corpora. When trained on English IR data with these losses and evaluated zero-shot on non-English data, our model demonstrates significant improvement to prior work on retrieval performance, while it requires much less computational effort. We also demonstrate the value of our model for a practical setting when a parallel corpus is only available for a few languages, but a lack of parallel corpora resources persists for many other low-resource languages. Our model can work well even with a small number of parallel sentences, and be used as an add-on module to any backbones and other tasks., Comment: ACL Findings 2023

Click-through rate (CTR) prediction is a crucial task in online display advertising. The embedding-based neural networks have been proposed to learn both explicit feature interactions through a shallow component and deep feature interactions using a deep neural network (DNN) component. These sophisticated models, however, slow down the prediction inference by at least hundreds of times. To address the issue of significantly increased serving delay and high memory usage for ad serving in production, this paper presents \emph{DeepLight}: a framework to accelerate the CTR predictions in three aspects: 1) accelerate the model inference via explicitly searching informative feature interactions in the shallow component; 2) prune redundant layers and parameters at intra-layer and inter-layer level in the DNN component; 3) promote the sparsity of the embedding layer to preserve the most discriminant signals. By combining the above efforts, the proposed approach accelerates the model inference by 46X on Criteo dataset and 27X on Avazu dataset without any loss on the prediction accuracy. This paves the way for successfully deploying complicated embedding-based neural networks in production for ad serving., Comment: Accepted by WSDM 2021; Source code: https://github.com/WayneDW/DeepLight_Deep-Lightweight-Feature-Interactions

Jianing Tang,1 Deguang Kong,2 Qiuxia Cui,1 Kun Wang,3 Dan Zhang,3 Xing Liao,1 Yan Gong,4 Gaosong Wu1 1Department of Thyroid and Breast Surgery, Zhongnan Hospital of Wuhan University, Wuhan, China; 2Department of General Surgery, Zhongnan Hospital of Wuhan University, Wuhan, China; 3Department of Breast and Thyroid Surgery, Tongji Hospital of Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China; 4Department of Biological Repositories, Zhongnan Hospital of Wuhan University, Wuhan, China Background: Papillary thyroid cancer (PTC) is a common endocrine malignancy with relatively good prognosis. Radioactive iodine (RAI) is considered effective for patients with total or nearly total thyroidectomy, but the beneficial effects of RAI are still controversial. Materials and methods: To determine whether RAI therapy could improve the survival rates of PTC patients, we conducted a retrospective analysis using data from the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) program. Disease-specific survival (DSS) was obtained using multivariate Cox proportional hazard regressions. Results: DSS was improved by RAI ablation in patients with tumor >2 cm, age >45 years and gross extrathyroidal or lymph node metastasis. In a further analysis, RAI therapy did not improve the DSS in patients with tumor <2 cm except those with distant metastasis. For patients with tumor >2 cm, those involving gross extrathyroidal extension, age >45 years or disease in the lymph nodes, DSS was improved after RAI therapy. Patients with distant metastasis always benefited from RAI ablation. Conclusion: RAI ablation should be recommended to patients with tumor <2 cm and distant metastasis or patients with tumor >2 cm and one of the following risk factors: gross extrathyroidal extension, age >45 years, lymph node and distant metastases. Keywords: RAI, prognosis, tumor size, metastasis

Deploying deep neural networks on mobile devices is a challenging task. Current model compression methods such as matrix decomposition effectively reduce the deployed model size, but still cannot satisfy real-time processing requirement. This paper first discovers that the major obstacle is the excessive execution time of non-tensor layers such as pooling and normalization without tensor-like trainable parameters. This motivates us to design a novel acceleration framework: DeepRebirth through "slimming" existing consecutive and parallel non-tensor and tensor layers. The layer slimming is executed at different substructures: (a) streamline slimming by merging the consecutive non-tensor and tensor layer vertically; (b) branch slimming by merging non-tensor and tensor branches horizontally. The proposed optimization operations significantly accelerate the model execution and also greatly reduce the run-time memory cost since the slimmed model architecture contains less hidden layers. To maximally avoid accuracy loss, the parameters in new generated layers are learned with layer-wise fine-tuning based on both theoretical analysis and empirical verification. As observed in the experiment, DeepRebirth achieves more than 3x speed-up and 2.5x run-time memory saving on GoogLeNet with only 0.4% drop of top-5 accuracy on ImageNet. Furthermore, by combining with other model compression techniques, DeepRebirth offers an average of 65ms inference time on the CPU of Samsung Galaxy S6 with 86.5% top-5 accuracy, 14% faster than SqueezeNet which only has a top-5 accuracy of 80.5%., Comment: AAAI 2018

Recent years have witnessed the increasing popularity of mobile devices (such as iphone) due to the convenience that it brings to human lives. On one hand, rich user profiling and behavior data (including per-app level, app-interaction level and system-interaction level) from heterogeneous information sources make it possible to provide much better services (such as recommendation, advertisement targeting) to customers, which further drives revenue from understanding users' behaviors and improving user' engagement. In order to delight the customers, intelligent personal assistants (such as Amazon Alexa, Google Home and Google Now) are highly desirable to provide real-time audio, video and image recognition, natural language understanding, comfortable user interaction interface, satisfactory recommendation and effective advertisement targeting. This paper presents the research efforts we have conducted on mobile devices which aim to provide much smarter and more convenient services by leveraging statistics and big data science, machine learning and deep learning, user modeling and marketing techniques to bring in significant user growth and user engagement and satisfactions (and happiness) on mobile devices. The developed new features are built at either cloud side or device side, harmonically working together to enhance the current service with the purpose of increasing users' happiness. We illustrate how we design these new features from system and algorithm perspective using different case studies, through which one can easily understand how science driven innovations help to provide much better service in technology and bring more revenue liftup in business. In the meantime, these research efforts have clear scientific contributions and published in top venues, which are playing more and more important roles for mobile AI products.

Although originally identified for its function in Drosophila melanogaster eye specification, the Retinal Determination Gene Network (RDGN) is essential for the development of multiple organs in mammals. The RDGN regulates proliferation, differentiation and autocrine signaling, and interacts with other key signaling pathways. Aberrant expression of RDGN members such as DACH, EYA and SIX contributes to tumor initiation and progression; indeed, the levels of RDGN members are clinically prognostic factors in various cancer types. Stimulation or suppression of the activities of these crucial components can block cancer cell proliferation, prevent cancer stem cell expansion and even reverse the EMT process, thereby attenuating malignant phenotypes. Thus, cancer therapeutic interventions targeting RDGN members should be pursued in future studies.

Local Linear embedding (LLE) is a popular dimension reduction method. In this paper, we first show LLE with nonnegative constraint is equivalent to the widely used Laplacian embedding. We further propose to iterate the two steps in LLE repeatedly to improve the results. Thirdly, we relax the kNN constraint of LLE and present a sparse similarity learning algorithm. The final Iterative LLE combines these three improvements. Extensive experiment results show that iterative LLE algorithm significantly improve both classification and clustering results., Comment: Appears in Proceedings of the 29th International Conference on Machine Learning (ICML 2012)