Fan Yang

Tao Shan, Wei Tang, Xunwang Dang et al.

In this work, we investigated the feasibility of applying deep learning techniques to solve Poisson's equation. A deep convolutional neural network is set up to predict the distribution of electric potential in 2D or 3D cases. With proper training data generated from a finite difference solver, the strong approximation capability of the deep convolutional neural network allows it to make correct prediction given information of the source and distribution of permittivity. With applications of L2 regularization, numerical experiments show that the predication error of 2D cases can reach below 1.5\% and the predication of 3D cases can reach below 3\%, with a significant reduction in CPU time compared with the traditional solver based on finite difference methods.

Elan Markowitz, Ziyan Jiang, Fan Yang et al. · amazon-science

This work explores unifying knowledge enhanced recommendation with multi-domain recommendation systems in a conversational AI assistant application. Multi-domain recommendation leverages users' interactions in previous domains to improve recommendations in a new one. Knowledge graph enhancement seeks to use external knowledge graphs to improve recommendations within a single domain. Both research threads incorporate related information to improve the recommendation task. We propose to unify these approaches: using information from interactions in other domains as well as external knowledge graphs to make predictions in a new domain that would not be possible with either information source alone. We develop a new model and demonstrate the additive benefit of these approaches on a dataset derived from millions of users' queries for content across three domains (videos, music, and books) in a live virtual assistant application. We demonstrate significant improvement on overall recommendations as well as on recommendations for new users of a domain.

Xiaowen Shi, Fan Yang, Ze Wang et al.

Re-ranking draws increased attention on both academics and industries, which rearranges the ranking list by modeling the mutual influence among items to better meet users' demands. Many existing re-ranking methods directly take the initial ranking list as input, and generate the optimal permutation through a well-designed context-wise model, which brings the evaluation-before-reranking problem. Meanwhile, evaluating all candidate permutations brings unacceptable computational costs in practice. Thus, to better balance efficiency and effectiveness, online systems usually use a two-stage architecture which uses some heuristic methods such as beam-search to generate a suitable amount of candidate permutations firstly, which are then fed into the evaluation model to get the optimal permutation. However, existing methods in both stages can be improved through the following aspects. As for generation stage, heuristic methods only use point-wise prediction scores and lack an effective judgment. As for evaluation stage, most existing context-wise evaluation models only consider the item context and lack more fine-grained feature context modeling. This paper presents a novel end-to-end re-ranking framework named PIER to tackle the above challenges which still follows the two-stage architecture and contains two mainly modules named FPSM and OCPM. We apply SimHash in FPSM to select top-K candidates from the full permutation based on user's permutation-level interest in an efficient way. Then we design a novel omnidirectional attention mechanism in OCPM to capture the context information in the permutation. Finally, we jointly train these two modules end-to-end by introducing a comparative learning loss. Offline experiment results demonstrate that PIER outperforms baseline models on both public and industrial datasets, and we have successfully deployed PIER on Meituan food delivery platform.

Gemini Team, Petko Georgiev, Ving Ian Lei et al. · deepmind, mila

In this report, we introduce the Gemini 1.5 family of models, representing the next generation of highly compute-efficient multimodal models capable of recalling and reasoning over fine-grained information from millions of tokens of context, including multiple long documents and hours of video and audio. The family includes two new models: (1) an updated Gemini 1.5 Pro, which exceeds the February version on the great majority of capabilities and benchmarks; (2) Gemini 1.5 Flash, a more lightweight variant designed for efficiency with minimal regression in quality. Gemini 1.5 models achieve near-perfect recall on long-context retrieval tasks across modalities, improve the state-of-the-art in long-document QA, long-video QA and long-context ASR, and match or surpass Gemini 1.0 Ultra's state-of-the-art performance across a broad set of benchmarks. Studying the limits of Gemini 1.5's long-context ability, we find continued improvement in next-token prediction and near-perfect retrieval (>99%) up to at least 10M tokens, a generational leap over existing models such as Claude 3.0 (200k) and GPT-4 Turbo (128k). Finally, we highlight real-world use cases, such as Gemini 1.5 collaborating with professionals on completing their tasks achieving 26 to 75% time savings across 10 different job categories, as well as surprising new capabilities of large language models at the frontier; when given a grammar manual for Kalamang, a language with fewer than 200 speakers worldwide, the model learns to translate English to Kalamang at a similar level to a person who learned from the same content.

Haiyan Zhao, Fan Yang, Bo Shen et al.

Large language models (LLMs) have led to breakthroughs in language tasks, yet the internal mechanisms that enable their remarkable generalization and reasoning abilities remain opaque. This lack of transparency presents challenges such as hallucinations, toxicity, and misalignment with human values, hindering the safe and beneficial deployment of LLMs. This paper aims to uncover the mechanisms underlying LLM functionality through the lens of explainability. First, we review how knowledge is architecturally composed within LLMs and encoded in their internal parameters via mechanistic interpretability techniques. Then, we summarize how knowledge is embedded in LLM representations by leveraging probing techniques and representation engineering. Additionally, we investigate the training dynamics through a mechanistic perspective to explain phenomena such as grokking and memorization. Lastly, we explore how the insights gained from these explanations can enhance LLM performance through model editing, improve efficiency through pruning, and better align with human values.

Youtan Yin, Zhoujie Fu, Fan Yang et al.

The emergence of Neural Radiance Fields (NeRF) for novel view synthesis has increased interest in 3D scene editing. An essential task in editing is removing objects from a scene while ensuring visual reasonability and multiview consistency. However, current methods face challenges such as time-consuming object labeling, limited capability to remove specific targets, and compromised rendering quality after removal. This paper proposes a novel object-removing pipeline, named OR-NeRF, that can remove objects from 3D scenes with user-given points or text prompts on a single view, achieving better performance in less time than previous works. Our method spreads user annotations to all views through 3D geometry and sparse correspondence, ensuring 3D consistency with less processing burden. Then recent 2D segmentation model Segment-Anything (SAM) is applied to predict masks, and a 2D inpainting model is used to generate color supervision. Finally, our algorithm applies depth supervision and perceptual loss to maintain consistency in geometry and appearance after object removal. Experimental results demonstrate that our method achieves better editing quality with less time than previous works, considering both quality and quantity.

Kishore Tumarada, Yifan Zhang, Fan Yang et al.

Opinion prediction is an emerging research area with diverse real-world applications, such as market research and situational awareness. We identify two lines of approaches to the problem of opinion prediction. One uses topic-based sentiment analysis with time-series modeling, while the other uses static embedding of text. The latter approaches seek user-specific solutions by generating user fingerprints. Such approaches are useful in predicting user's reactions to unseen content. In this work, we propose a novel dynamic fingerprinting method that leverages contextual embedding of user's comments conditioned on relevant user's reading history. We integrate BERT variants with a recurrent neural network to generate predictions. The results show up to 13\% improvement in micro F1-score compared to previous approaches. Experimental results show novel insights that were previously unknown such as better predictions for an increase in dynamic history length, the impact of the nature of the article on performance, thereby laying the foundation for further research.

Yifan Zhang, Dainis Boumber, Marjan Hosseinia et al.

We propose an unsupervised solution to the Authorship Verification task that utilizes pre-trained deep language models to compute a new metric called DV-Distance. The proposed metric is a measure of the difference between the two authors comparing against pre-trained language models. Our design addresses the problem of non-comparability in authorship verification, frequently encountered in small or cross-domain corpora. To the best of our knowledge, this paper is the first one to introduce a method designed with non-comparability in mind from the ground up, rather than indirectly. It is also one of the first to use Deep Language Models in this setting. The approach is intuitive, and it is easy to understand and interpret through visualization. Experiments on four datasets show our methods matching or surpassing current state-of-the-art and strong baselines in most tasks.

Yifan Zhang, Fan Yang, Marjan Hosseinia et al.

In this paper, we introduce a new framework called the sentiment-aspect attribution module (SAAM). SAAM works on top of traditional neural networks and is designed to address the problem of multi-aspect sentiment classification and sentiment regression. The framework works by exploiting the correlations between sentence-level embedding features and variations of document-level aspect rating scores. We demonstrate several variations of our framework on top of CNN and RNN based models. Experiments on a hotel review dataset and a beer review dataset have shown SAAM can improve sentiment analysis performance over corresponding base models. Moreover, because of the way our framework intuitively combines sentence-level scores into document-level scores, it is able to provide a deeper insight into data (e.g., semi-supervised sentence aspect labeling). Hence, we end the paper with a detailed analysis that shows the potential of our models for other applications such as sentiment snippet extraction.

Wen Dong, Bo Liu, Fan Yang

Complex social systems are composed of interconnected individuals whose interactions result in group behaviors. Optimal control of a real-world complex system has many applications, including road traffic management, epidemic prevention, and information dissemination. However, such real-world complex system control is difficult to achieve because of high-dimensional and non-linear system dynamics, and the exploding state and action spaces for the decision maker. Prior methods can be divided into two categories: simulation-based and analytical approaches. Existing simulation approaches have high-variance in Monte Carlo integration, and the analytical approaches suffer from modeling inaccuracy. We adopted simulation modeling in specifying the complex dynamics of a complex system, and developed analytical solutions for searching optimal strategies in a complex network with high-dimensional state-action space. To capture the complex system dynamics, we formulate the complex social network decision making problem as a discrete event decision process. To address the curse of dimensionality and search in high-dimensional state action spaces in complex systems, we reduce control of a complex system to variational inference and parameter learning, introduce Bethe entropy approximation, and develop an expectation propagation algorithm. Our proposed algorithm leads to higher system expected rewards, faster convergence, and lower variance of value function in a real-world transportation scenario than state-of-the-art analytical and sampling approaches.

Fan Yang, Santiago Escobar, Catherine Meadows et al.

Roles in cryptographic protocols do not always have a linear execution, but may include choice points causing the protocol to continue along different paths. In this paper we address the problem of representing choice in the strand space model of cryptographic protocols, particularly as it is used in the Maude-NPA cryptographic protocol analysis tool. To achieve this goal, we develop and give formal semantics to a process algebra for cryptographic protocols that supports a rich taxonomy of choice primitives for composing strand spaces. In our taxonomy, deterministic and non-deterministic choices are broken down further. Non-deterministic choice can be either explicit, i.e., one of two paths is chosen, or implicit, i.e., the value of a variable is chosen non-deterministically. Likewise, deterministic choice can be either an explicit if-then-else choice, i.e., one path is chosen if a predicate is satisfied, while the other is chosen if it is not, or implicit deterministic choice, i.e., execution continues only if a certain pattern is matched. We have identified a class of choices which includes finite branching and some cases of infinite branching, which we address in this paper. We provide a bisimulation result between the expected forwards execution semantics of the new process algebra and the original symbolic backwards semantics of Maude-NPA that preserves attack reachability. We have fully integrated the process algebra syntax and its transformation into strands in Maude-NPA. We illustrate its expressive power and naturalness with various examples, and show how it can be effectively used in formal analysis. This allows users to write protocols from now on using the process syntax, which is more convenient for expressing choice than the strand space syntax, in which choice can only be specified implicitly, via two or more strands that are identical until the choice point.

Fan Yang, Arjun Mukherjee, Eduard Dragut

Satirical news is considered to be entertainment, but it is potentially deceptive and harmful. Despite the embedded genre in the article, not everyone can recognize the satirical cues and therefore believe the news as true news. We observe that satirical cues are often reflected in certain paragraphs rather than the whole document. Existing works only consider document-level features to detect the satire, which could be limited. We consider paragraph-level linguistic features to unveil the satire by incorporating neural network and attention mechanism. We investigate the difference between paragraph-level features and document-level features, and analyze them on a large satirical news dataset. The evaluation shows that the proposed model detects satirical news effectively and reveals what features are important at which level.