United States

Deep neural networks (DNNs) have been widely adopted in many decision-making industrial applications. Their fairness issues, i.e., whether there exist unintended biases in the DNN, receive much attention and become critical concerns, which can directly cause negative impacts in our daily life and potentially undermine the fairness of our society, especially with their increasing deployment at an unprecedented speed. Recently, some early attempts have been made to provide fairness assurance of DNNs, such as fairness testing, which aims at finding discriminatory samples empirically, and fairness certification, which develops sound but not complete analysis to certify the fairness of DNNs. Nevertheless, how to formally compute discriminatory samples and fairness scores (i.e., the percentage of fair input space), is still largely uninvestigated. In this paper, we propose DeepGemini, a novel fairness formal analysis technique for DNNs, which contains two key components: discriminatory sample discovery and fairness score computation. To uncover discriminatory samples, we encode the fairness of DNNs as safety properties and search for discriminatory samples by means of state-of-the-art verification techniques for DNNs. This reduction enables us to be the first to formally compute discriminatory samples. To compute the fairness score, we develop counterexample guided fairness analysis, which utilizes four heuristics to efficiently approximate a lower bound of fairness score. Extensive experimental evaluations demonstrate the effectiveness and efficiency of DeepGemini on commonly-used benchmarks, and DeepGemini outperforms state-of-the-art DNN fairness certification approaches in terms of both efficiency and scalability.

AAAI 2023

DeepGemini: Verifying Dependency Fairness for Deep Neural Network

quality assurance of dnn

fairness verification

neural network verification

The purpose of the AAAI conference series is to promote research in Artificial Intelligence (AI) and foster scientific exchange between researchers, practitioners, scientists, students, and engineers across the entirety of AI and its affiliated disciplines. AAAI-23 is the Thirty-Seventh AAAI Conference on Artificial Intelligence. The theme of this conference is to create collaborative bridges within and beyond AI. Like previous AAAI conferences, AAAI-23 will feature technical paper presentations, special tracks, invited speakers, workshops, tutorials, poster sessions, senior member presentations, competitions, and exhibit programs, and two new activities: a Bridge Program and a Lab Program. Many of these activities are tailored to the theme of bridges and all are selected according to the highest standards, with additional programs for students and young researchers. 
AAAI is providing you with a conference planner, which you can use to help organize your itinerary of activities. This includes talks to attend in person, talks to attend remotely, breaks with colleagues and your site seeing activities. To access this conference planner, please go to [https://aaai-2023.takemobi.io](https://aaai-2023.takemobi.io).

In order to access this site, you need to register. If you haven't already, please register [here](https://aaai.org/Conferences/AAAI-23/registration/).


The purpose of the AAAI conference series is to promote research in Artificial Intelligence (AI) and foster scientific exchange between researchers, practitioners, scientists, students, and engineers across the entirety of AI and its affiliated disciplines.

poster

Nowadays, AI-based techniques, such as deep neural networks (DNNs), are widely deployed in autonomous systems for complex mission requirements (e.g., motion planning in robotics). However, DNNs-based controllers are typically very complex, and it is very hard to formally verify their correctness, potentially causing severe risks for safety-critical autonomous systems. In this paper, we propose a construction scheme for a so-called Safe-visor architecture to sandbox DNNs-based controllers. Particularly, we consider the construction under a stochastic game framework to provide a system-level safety guarantee which is robust to noises and disturbances. A supervisor is built to check the control inputs provided by a DNNs-based controller and decide whether to accept them.  Meanwhile, a safety advisor is running in parallel to provide fallback control inputs in case the DNN-based controller is rejected. We demonstrate the proposed approaches on a quadrotor employing an unverified DNNs-based controller.

Towards Safe AI: Sandboxing DNNs-based Controllers in Stochastic Games

A family of methods that generate soft labels by mixing the hard labels with a certain distribution, namely label refurbishment, are widely used to train deep neural networks. However, some of these methods are still poorly understood in the presence of label noise. In this paper, we revisit four label refurbishment methods and reveal the strong connection between them. We find that they affect the neural network models in different manners. Two of them smooth the estimated posterior for regularization effects, and the other two force the model to produce high-confidence predictions. We conduct extensive experiments to evaluate related methods and observe that both effects improve the model generalization under label noise. Furthermore, we theoretically show that both effects lead to generalization guarantees on the clean distribution despite being trained with noisy labels.

Rethinking Label Refurbishment: Model Robustness under Label Noise

We formally verify executable algorithms for solving Markov decision processes (MDPs) in the interactive theorem prover Isabelle/HOL. We build on existing formalizations of probability theory to analyze the expected total reward criterion on finite and infinite-horizon problems. Our developments formalize the Bellman equation and give conditions under which optimal policies exist. Based on this analysis, we verify dynamic programming algorithms to solve tabular MDPs. We evaluate the formally verified implementations experimentally on standard problems, compare them with state-of-the-art systems, and show that they are practical.

Formally Verified Solution Methods for Markov Decision Processes

Spatio-temporal machine learning is critically needed for a variety of societal applications, such as agricultural monitoring, hydrological forecast, and traffic management. These applications greatly rely on regional features that characterize spatial and temporal differences. However, spatio-temporal data are often complex and pose several unique challenges for machine learning models: 1) multiple models are needed to handle region-based data patterns that have significant spatial heterogeneity across different locations; 2) local models trained on region-specific data have limited ability to adapt to other regions that have large diversity and abnormality; 3) spatial and temporal variations entangle data complexity that requires more robust and adaptive models; 4) limited spatial-temporal data in real scenarios (e.g., crop yield data is collected only once a year) makes the problems intrinsically challenging. To bridge these gaps, we propose task-adaptive formulations and a model-agnostic meta-learning framework that ensembles regionally heterogeneous data into location-sensitive meta tasks. We conduct task adaptation following an easy-to-hard task hierarchy in which different meta models are adapted to tasks of different difficulty levels. One major advantage of our proposed method is that it improves the model adaptation to a large number of heterogeneous tasks. It also enhances the model generalization by automatically adapting the meta model of the corresponding difficulty level to any new tasks. We demonstrate the superiority of our proposed framework over a diverse set of baselines and state-of-the-art meta-learning frameworks. Our extensive experiments on real crop yield data show the effectiveness of the proposed method in handling spatial-related heterogeneous tasks in real societal applications.

Task-Adaptive Meta-Learning Framework for Advancing Spatial Generalizability

Although significant progress has been made in face recognition, demographic bias still exists in face recognition systems. For instance, it usually happens that the face recognition performance for a certain demographic group is lower than the others. In this paper, we propose MixFairFace framework to improve the fairness in face recognition models. First of all, we argue that the commonly used attribute-based fairness metric is not appropriate for face recognition. A face recognition system can only be considered fair while every person has a close performance. Hence, we propose a new evaluation protocol to fairly evaluate the fairness performance of different approaches. Different from previous approaches that require sensitive attribute labels such as race and gender for reducing the demographic bias, we aim at addressing the identity bias in face representation, i.e., the performance inconsistency between different identities, without the need for sensitive attribute labels. To this end, we propose MixFair Adapter to determine and reduce the identity bias of training samples. Our extensive experiments demonstrate that our MixFairFace approach achieves state-of-the-art fairness performance on all benchmark datasets.

MixFairFace: Towards Ultimate Fairness via MixFair Adapter in Face Recognition

Automatic Text Scoring (ATS) is a widely-investigated task in education. Existing approaches often stressed the structure design of an ATS model and neglected the training process of the model. Considering the difficult nature of this task, we argued that the performance of an ATS model could be potentially boosted by carefully selecting data of varying complexities in the training process. Therefore, we aimed to investigate the effectiveness of curriculum learning (CL) in scoring educational text. Specifically, we designed two types of difficulty measurers: (i) pre-defined, calculated by measuring a sample's readability, length, the number of grammatical errors or unique words it contains; and (ii) automatic, calculated based on whether a model in a training epoch can accurately score the samples. These measurers were tested in both the easy-to-hard to hard-to-easy training paradigms. Through extensive evaluations on two widely-used datasets (one for short answer scoring and the other for long essay scoring), we demonstrated that (a) CL indeed could boost the performance of state-of-the-art ATS models, and the maximum improvement could be up to 4.5%, but most improvements were achieved when assessing short and easy answers; (b) the pre-defined measurer calculated based on the number of grammatical errors contained in a text sample tended to outperform the other difficulty measurers across different training paradigms.

On the Effectiveness of Curriculum Learning in Educational Text Scoring

There has been increasing concern within the machine learning community and beyond that Artificial Intelligence (AI) faces a bias and discrimination crisis which needs AI fairness with urgency. As many have begun to work on this problem, most existing work depends on the availability of class label for the given fairness definition and algorithm which may not align with real-world usage. In this work, we study an AI fairness problem that stems from the gap between the design of a ``fair'' model in the lab and its deployment in the real-world. Specifically, we consider defining and mitigating individual unfairness as a survival analysis problem, where the availability of class label is not always guaranteed due to censorship, which is broadly applicable in a diversity of real-world, socially-sensitive applications. We show that our method is able to quantify and mitigate individual unfairness in the presence of censorship across four benchmark tasks, which provides the first known results on individual fairness guarantee in analysis of censored data.

Censored Fairness through Awareness

A significant cause of air pollution in urban areas worldwide is the high volume of road traffic. Long-term exposure to severe pollution can cause serious health issues. One approach towards tackling this problem is to design a pollution-aware traffic routing policy that balances multiple objectives of i) avoiding extreme pollution in any area ii) enabling short transit times, and iii) making effective use of the road capacities. We propose a novel sampling-based approach for this problem. We give the first construction of a Markov Chain that can sample integer max flow solutions of a planar graph, with theoretical guarantees that the probabilities depend on the aggregate transit length. We designed a traffic policy using diverse samples and simulated traffic on real-world road maps using the SUMO traffic simulator. We observe a considerable decrease in areas with severe pollution when experimented with maps of large cities across the world compared to other approaches. 

City-scale Pollution Aware Traffic Routing by Sampling Max Flows using MCMC

Recently, short video platforms have become an important channel for news sharing, but also a new breeding ground for fake news. To mitigate this problem, the research of fake news video detection has recently received a lot of attention. Existing works face two roadblocks:  the scarcity of comprehensive and large-scale datasets and insufficient utilization of multimodal information. Therefore, in this paper, we construct the largest Chinese short video dataset about fake news named FakeSV, which includes the news content, user comments, and user profiles simultaneously. To understand the characteristics of fake news videos, we conduct exploratory analyses of FakeSV from different perspectives. Moreover, we provide a new multimodal detection model named SV-FEND, which exploits the cross-modal correlations to select the most informative features and utilizes the social context information for detection. Extensive experiments evaluate the superiority of the proposed method and also provide detailed comparisons of different methods and modalities for future works.

FakeSV: A Multimodal Benchmark with Rich Social Context for Fake News Detection on Short Video Platforms

Epidemic models are powerful tools in understanding infectious disease. However, as they increase in size and complexity, they can quickly become computationally intractable. Recent progress in modelling methodology has shown that surrogate models can be used to emulate complex epidemic models with a high-dimensional parameter space. We show that deep sequence-to-sequence (seq2seq) models can serve as accurate surrogates for complex epidemic models with sequence based model parameters, effectively replicating seasonal and long-term transmission dynamics. Once trained, our surrogate can predict scenarios a several thousand times faster than the original model, making them ideal for policy exploration. We demonstrate that replacing a traditional epidemic model with a learned simulator facilitates robust Bayesian inference.

DeepGemini: Verifying Dependency Fairness for Deep Neural Network

Downloads

Next from AAAI 2023

Towards Safe AI: Sandboxing DNNs-based Controllers in Stochastic Games

Similar lecture

Learning Dynamic Latent Spaces for Lifelong Generative Modelling