Dong Lin

Rohan Anil, Sandra Gadanho, Da Huang et al. · deepmind

For industrial-scale advertising systems, prediction of ad click-through rate (CTR) is a central problem. Ad clicks constitute a significant class of user engagements and are often used as the primary signal for the usefulness of ads to users. Additionally, in cost-per-click advertising systems where advertisers are charged per click, click rate expectations feed directly into value estimation. Accordingly, CTR model development is a significant investment for most Internet advertising companies. Engineering for such problems requires many machine learning (ML) techniques suited to online learning that go well beyond traditional accuracy improvements, especially concerning efficiency, reproducibility, calibration, credit attribution. We present a case study of practical techniques deployed in Google's search ads CTR model. This paper provides an industry case study highlighting important areas of current ML research and illustrating how impactful new ML methods are evaluated and made useful in a large-scale industrial setting.

Le Yan, Zhen Qin, Gil Shamir et al.

Graded labels are ubiquitous in real-world learning-to-rank applications, especially in human rated relevance data. Traditional learning-to-rank techniques aim to optimize the ranked order of documents. They typically, however, ignore predicting actual grades. This prevents them from being adopted in applications where grades matter, such as filtering out ``poor'' documents. Achieving both good ranking performance and good grade prediction performance is still an under-explored problem. Existing research either focuses only on ranking performance by not calibrating model outputs, or treats grades as numerical values, assuming labels are on a linear scale and failing to leverage the ordinal grade information. In this paper, we conduct a rigorous study of learning to rank with grades, where both ranking performance and grade prediction performance are important. We provide a formal discussion on how to perform ranking with non-scalar predictions for grades, and propose a multiobjective formulation to jointly optimize both ranking and grade predictions. In experiments, we verify on several public datasets that our methods are able to push the Pareto frontier of the tradeoff between ranking and grade prediction performance, showing the benefit of leveraging ordinal grade information.

Roy Xie, Deepak Gopinath, David Qiu et al.

Search-augmented large language models (LLMs) excel at knowledge-intensive tasks by integrating external retrieval. However, they often over-search -- unnecessarily invoking search tool even when it does not improve response quality, which leads to computational inefficiency and hallucinations by incorporating irrelevant context. In this work, we conduct a systematic evaluation of over-searching across multiple dimensions, including query types, model categories, retrieval conditions, and multi-turn conversations. Our finding shows: (i) search generally improves answer accuracy on answerable queries but harms abstention on unanswerable ones; (ii) over-searching is more pronounced in complex reasoning models and deep research systems, is exacerbated by noisy retrieval, and compounds across turns in multi-turn conversations; and (iii) the composition of retrieved evidence is crucial, as the presence of negative evidence improves abstention. To quantify over-searching, we introduce Tokens Per Correctness (TPC), an evaluation metric that captures the performance-cost trade-off for search-augmented LLMs. Lastly, we investigate mitigation approaches at both the query and retrieval levels and release the OverSearchQA to foster continued research into efficient search-augmented LLMs.

Lei Ding, Kai Zheng, Dong Lin et al.

There are limited studies on the semantic segmentation of high-resolution Polarimetric Synthetic Aperture Radar (PolSAR) images due to the scarcity of training data and the inference of speckle noises. The Gaofen contest has provided open access of a high-quality PolSAR semantic segmentation dataset. Taking this chance, we propose a Multi-path ResNet (MP-ResNet) architecture for the semantic segmentation of high-resolution PolSAR images. Compared to conventional U-shape encoder-decoder convolutional neural network (CNN) architectures, the MP-ResNet learns semantic context with its parallel multi-scale branches, which greatly enlarges its valid receptive fields and improves the embedding of local discriminative features. In addition, MP-ResNet adopts a multi-level feature fusion design in its decoder to make the best use of the features learned from its different branches. Ablation studies show that the MPResNet has significant advantages over its baseline method (FCN with ResNet34). It also surpasses several classic state-of-the-art methods in terms of overall accuracy (OA), mean F1 and fwIoU, whereas its computational costs are not much increased. This CNN architecture can be used as a baseline method for future studies on the semantic segmentation of PolSAR images. The code is available at: https://github.com/ggsDing/SARSeg.

Roy Xie, David Qiu, Deepak Gopinath et al.

Long chain-of-thought (CoT) significantly enhances large language models' (LLM) reasoning capabilities. However, the extensive reasoning traces lead to inefficiencies and an increased time-to-first-token (TTFT). We propose a novel training paradigm that uses reinforcement learning (RL) to guide reasoning LLMs to interleave thinking and answering for multi-hop questions. We observe that models inherently possess the ability to perform interleaved reasoning, which can be further enhanced through RL. We introduce a simple yet effective rule-based reward to incentivize correct intermediate steps, which guides the policy model toward correct reasoning paths by leveraging intermediate signals generated during interleaved reasoning. Extensive experiments conducted across five diverse datasets and three RL algorithms (PPO, GRPO, and REINFORCE++) demonstrate consistent improvements over traditional think-answer reasoning, without requiring external tools. Specifically, our approach reduces TTFT by over 80% on average and improves up to 19.3% in Pass@1 accuracy. Furthermore, our method, trained solely on question answering and logical reasoning datasets, exhibits strong generalization ability to complex reasoning datasets such as MATH, GPQA, and MMLU. Additionally, we conduct in-depth analysis to reveal several valuable insights into conditional reward modeling.

Gil I. Shamir, Dong Lin

Real world recommendation systems influence a constantly growing set of domains. With deep networks, that now drive such systems, recommendations have been more relevant to the user's interests and tasks. However, they may not always be reproducible even if produced by the same system for the same user, recommendation sequence, request, or query. This problem received almost no attention in academic publications, but is, in fact, very realistic and critical in real production systems. We consider reproducibility of real large scale deep models, whose predictions determine such recommendations. We demonstrate that the celebrated Rectified Linear Unit (ReLU) activation, used in deep models, can be a major contributor to irreproducibility. We propose the use of smooth activations to improve recommendation reproducibility. We describe a novel family of smooth activations; Smooth ReLU (SmeLU), designed to improve reproducibility with mathematical simplicity, with potentially cheaper implementation. SmeLU is a member of a wider family of smooth activations. While other techniques that improve reproducibility in real systems usually come at accuracy costs, smooth activations not only improve reproducibility, but can even give accuracy gains. We report metrics from real systems in which we were able to productionalize SmeLU with substantial reproducibility gains and better accuracy-reproducibility trade-offs. These include click-through-rate (CTR) prediction systems, content, and application recommendation systems.

Haichao Yu, Zhe Chen, Dong Lin et al.

Dropout has been commonly used to quantify prediction uncertainty, i.e, the variations of model predictions on a given input example. However, using dropout in practice can be expensive as it requires running dropout inferences many times. In this paper, we study how to estimate dropout prediction uncertainty in a resource-efficient manner. We demonstrate that we can use neuron activation strengths to estimate dropout prediction uncertainty under different dropout settings and on a variety of tasks using three large datasets, MovieLens, Criteo, and EMNIST. Our approach provides an inference-once method to estimate dropout prediction uncertainty as a cheap auxiliary task. We also demonstrate that using activation features from a subset of the neural network layers can be sufficient to achieve uncertainty estimation performance almost comparable to that of using activation features from all layers, thus reducing resources even further for uncertainty estimation.

Lei Ding, Dong Lin, Shaofu Lin et al.

Long-range contextual information is crucial for the semantic segmentation of High-Resolution (HR) Remote Sensing Images (RSIs). However, image cropping operations, commonly used for training neural networks, limit the perception of long-range contexts in large RSIs. To overcome this limitation, we propose a Wide-Context Network (WiCoNet) for the semantic segmentation of HR RSIs. Apart from extracting local features with a conventional CNN, the WiCoNet has an extra context branch to aggregate information from a larger image area. Moreover, we introduce a Context Transformer to embed contextual information from the context branch and selectively project it onto the local features. The Context Transformer extends the Vision Transformer, an emerging kind of neural network, to model the dual-branch semantic correlations. It overcomes the locality limitation of CNNs and enables the WiCoNet to see the bigger picture before segmenting the land-cover/land-use (LCLU) classes. Ablation studies and comparative experiments conducted on several benchmark datasets demonstrate the effectiveness of the proposed method. In addition, we present a new Beijing Land-Use (BLU) dataset. This is a large-scale HR satellite dataset with high-quality and fine-grained reference labels, which can facilitate future studies in this field.

Gil I. Shamir, Dong Lin, Lorenzo Coviello

Deep networks are gradually penetrating almost every domain in our lives due to their amazing success. However, with substantive performance accuracy improvements comes the price of \emph{irreproducibility}. Two identical models, trained on the exact same training dataset may exhibit large differences in predictions on individual examples even when average accuracy is similar, especially when trained on highly distributed parallel systems. The popular Rectified Linear Unit (ReLU) activation has been key to recent success of deep networks. We demonstrate, however, that ReLU is also a catalyzer to irreproducibility in deep networks. We show that not only can activations smoother than ReLU provide better accuracy, but they can also provide better accuracy-reproducibility tradeoffs. We propose a new family of activations; Smooth ReLU (\emph{SmeLU}), designed to give such better tradeoffs, while also keeping the mathematical expression simple, and thus implementation cheap. SmeLU is monotonic, mimics ReLU, while providing continuous gradients, yielding better reproducibility. We generalize SmeLU to give even more flexibility and then demonstrate that SmeLU and its generalized form are special cases of a more general methodology of REctified Smooth Continuous Unit (RESCU) activations. Empirical results demonstrate the superior accuracy-reproducibility tradeoffs with smooth activations, SmeLU in particular.

Ruoxi Wang, Rakesh Shivanna, Derek Z. Cheng et al.

Learning effective feature crosses is the key behind building recommender systems. However, the sparse and large feature space requires exhaustive search to identify effective crosses. Deep & Cross Network (DCN) was proposed to automatically and efficiently learn bounded-degree predictive feature interactions. Unfortunately, in models that serve web-scale traffic with billions of training examples, DCN showed limited expressiveness in its cross network at learning more predictive feature interactions. Despite significant research progress made, many deep learning models in production still rely on traditional feed-forward neural networks to learn feature crosses inefficiently. In light of the pros/cons of DCN and existing feature interaction learning approaches, we propose an improved framework DCN-V2 to make DCN more practical in large-scale industrial settings. In a comprehensive experimental study with extensive hyper-parameter search and model tuning, we observed that DCN-V2 approaches outperform all the state-of-the-art algorithms on popular benchmark datasets. The improved DCN-V2 is more expressive yet remains cost efficient at feature interaction learning, especially when coupled with a mixture of low-rank architecture. DCN-V2 is simple, can be easily adopted as building blocks, and has delivered significant offline accuracy and online business metrics gains across many web-scale learning to rank systems at Google.

Zhe Chen, Yuyan Wang, Dong Lin et al.

Despite deep neural network (DNN)'s impressive prediction performance in various domains, it is well known now that a set of DNN models trained with the same model specification and the same data can produce very different prediction results. Ensemble method is one state-of-the-art benchmark for prediction uncertainty estimation. However, ensembles are expensive to train and serve for web-scale traffic. In this paper, we seek to advance the understanding of prediction variation estimated by the ensemble method. Through empirical experiments on two widely used benchmark datasets MovieLens and Criteo in recommender systems, we observe that prediction variations come from various randomness sources, including training data shuffling, and parameter random initialization. By introducing more randomness into model training, we notice that ensemble's mean predictions tend to be more accurate while the prediction variations tend to be higher. Moreover, we propose to infer prediction variation from neuron activation strength and demonstrate the strong prediction power from activation strength features. Our experiment results show that the average R squared on MovieLens is as high as 0.56 and on Criteo is 0.81. Our method performs especially well when detecting the lowest and highest variation buckets, with 0.92 AUC and 0.89 AUC respectively. Our approach provides a simple way for prediction variation estimation, which opens up new opportunities for future work in many interesting areas (e.g.,model-based reinforcement learning) without relying on serving expensive ensemble models.

Yuyan Wang, Zhe Zhao, Bo Dai et al.

Multi-task learning aims at solving multiple machine learning tasks at the same time. A good solution to a multi-task learning problem should be generalizable in addition to being Pareto optimal. In this paper, we provide some insights on understanding the trade-off between Pareto efficiency and generalization as a result of parameterization in multi-task deep learning models. As a multi-objective optimization problem, enough parameterization is needed for handling task conflicts in a constrained solution space; however, from a multi-task generalization perspective, over-parameterization undermines the benefit of learning a shared representation which helps harder tasks or tasks with limited training examples. A delicate balance between multi-task generalization and multi-objective optimization is therefore needed for finding a better trade-off between efficiency and generalization. To this end, we propose a method of under-parameterized self-auxiliaries for multi-task models to achieve the best of both worlds. It is task-agnostic and works with other multi-task learning algorithms. Empirical results show that small towers of under-parameterized self-auxiliaries can make big differences in improving Pareto efficiency in various multi-task applications.

Wang-Cheng Kang, Derek Zhiyuan Cheng, Ting Chen et al.

Recommender system models often represent various sparse features like users, items, and categorical features via embeddings. A standard approach is to map each unique feature value to an embedding vector. The size of the produced embedding table grows linearly with the size of the vocabulary. Therefore, a large vocabulary inevitably leads to a gigantic embedding table, creating two severe problems: (i) making model serving intractable in resource-constrained environments; (ii) causing overfitting problems. In this paper, we seek to learn highly compact embeddings for large-vocab sparse features in recommender systems (recsys). First, we show that the novel Differentiable Product Quantization (DPQ) approach can generalize to recsys problems. In addition, to better handle the power-law data distribution commonly seen in recsys, we propose a Multi-Granular Quantized Embeddings (MGQE) technique which learns more compact embeddings for infrequent items. We seek to provide a new angle to improve recommendation performance with compact model sizes. Extensive experiments on three recommendation tasks and two datasets show that we can achieve on par or better performance, with only ~20% of the original model size.

Jiaxi Tang, Rakesh Shivanna, Zhe Zhao et al.

Knowledge Distillation (KD) is a model-agnostic technique to improve model quality while having a fixed capacity budget. It is a commonly used technique for model compression, where a larger capacity teacher model with better quality is used to train a more compact student model with better inference efficiency. Through distillation, one hopes to benefit from student's compactness, without sacrificing too much on model quality. Despite the large success of knowledge distillation, better understanding of how it benefits student model's training dynamics remains under-explored. In this paper, we categorize teacher's knowledge into three hierarchical levels and study its effects on knowledge distillation: (1) knowledge of the `universe', where KD brings a regularization effect through label smoothing; (2) domain knowledge, where teacher injects class relationships prior to student's logit layer geometry; and (3) instance specific knowledge, where teacher rescales student model's per-instance gradients based on its measurement on the event difficulty. Using systematic analyses and extensive empirical studies on both synthetic and real-world datasets, we confirm that the aforementioned three factors play a major role in knowledge distillation. Furthermore, based on our findings, we diagnose some of the failure cases of applying KD from recent studies.