Research Papers

Verifiable Fairness: Privacy-preserving Computation of Fairness for Machine Learning Systems

Ehsan Toreini, Maryam Mehrnezhad, Aad van Moorsel

Fair machine learning is a thriving and vibrant research topic. In this paper, we propose Fairness as a Service (FaaS), a secure, verifiable and privacy-preserving protocol to computes and verify the fairness of any machine learning (ML) model. In the deisgn of FaaS, the data and outcomes are represented through cryptograms to ensure privacy. Also, zero knowledge proofs guarantee the well-formedness of the cryptograms and underlying data. FaaS is model--agnostic and can support various fairness metrics; hence, it can be used as a service to audit the fairness of any ML model. Our solution requires no trusted third party or private channels for the computation of the fairness metric. The security guarantees and commitments are implemented in a way that every step is securely transparent and verifiable from the start to the end of the process. The cryptograms of all input data are publicly available for everyone, e.g., auditors, social activists and experts, to verify the correctness of the process. We implemented FaaS to investigate performance and demonstrate the successful use of FaaS for a publicly available data set with thousands of entries.

We address the task of probabilistic anomaly attribution in the black-box regression setting, where the goal is to compute the probability distribution of the attribution score of each input variable, given an observed anomaly. The training dataset is assumed to be unavailable. This task differs from the standard XAI (explainable AI) scenario, since we wish to explain the anomalous deviation from a black-box prediction rather than the black-box model itself. We begin by showing that mainstream model-agnostic explanation methods, such as the Shapley values, are not suitable for this task because of their ``deviation-agnostic property.'' We then propose a novel framework for probabilistic anomaly attribution that allows us to not only compute attribution scores as the predictive mean but also quantify the uncertainty of those scores. This is done by considering a generative process for perturbations that counter-factually bring the observed anomalous observation back to normalcy. We introduce a variational Bayes algorithm for deriving the distributions of per variable attribution scores. To the best of our knowledge, this is the first probabilistic anomaly attribution framework that is free from being deviation-agnostic.

The increasing usage of new data sources and machine learning (ML) technology in credit modeling raises concerns with regards to potentially unfair decision-making that rely on protected characteristics (e.g., race, sex, age) or other socio-economic and demographic data. The authors demonstrate the impact of such algorithmic bias in the microfinance context. Difficulties in assessing credit are disproportionately experienced among vulnerable groups, however, very little is known about inequities in credit allocation between groups defined, not only by single, but by multiple and intersecting social categories. Drawing from the intersectionality paradigm, the study examines intersectional horizontal inequities in credit access by gender, age, marital status, single parent status and number of children. This paper utilizes data from the Spanish microfinance market as its context to demonstrate how pluralistic realities and intersectional identities can shape patterns of credit allocation when using automated decision-making systems. With ML technology being oblivious to societal good or bad, we find that a more thorough examination of intersectionality can enhance the algorithmic fairness lens to more authentically empower action for equitable outcomes and present a fairer path forward. We demonstrate that while on a high-level, fairness may exist superficially, unfairness can exacerbate at lower levels given combinatorial effects; in other words, the core fairness problem may be more complicated than current literature demonstrates. We find that in addition to legally protected characteristics, sensitive attributes such as single parent status and number of children can result in imbalanced harm. We discuss the implications of these findings for the financial services industry.

HiTZ@Antidote: Argumentation-driven Explainable Artificial Intelligence for Digital Medicine

Rodrigo Agerri, Iñigo Alonso, Aitziber Atutxa +9

Providing high quality explanations for AI predictions based on machine learning is a challenging and complex task. To work well it requires, among other factors: selecting a proper level of generality/specificity of the explanation; considering assumptions about the familiarity of the explanation beneficiary with the AI task under consideration; referring to specific elements that have contributed to the decision; making use of additional knowledge (e.g. expert evidence) which might not be part of the prediction process; and providing evidence supporting negative hypothesis. Finally, the system needs to formulate the explanation in a clearly interpretable, and possibly convincing, way. Given these considerations, ANTIDOTE fosters an integrated vision of explainable AI, where low-level characteristics of the deep learning process are combined with higher level schemes proper of the human argumentation capacity. ANTIDOTE will exploit cross-disciplinary competences in deep learning and argumentation to support a broader and innovative view of explainable AI, where the need for high-quality explanations for clinical cases deliberation is critical. As a first result of the project, we publish the Antidote CasiMedicos dataset to facilitate research on explainable AI in general, and argumentation in the medical domain in particular.

A recent trend in explainable AI research has focused on surrogate modeling, where neural networks are approximated as simpler ML algorithms such as kernel machines. A second trend has been to utilize kernel functions in various explain-by-example or data attribution tasks. In this work, we combine these two trends to analyze approximate empirical neural tangent kernels (eNTK) for data attribution. Approximation is critical for eNTK analysis due to the high computational cost to compute the eNTK. We define new approximate eNTK and perform novel analysis on how well the resulting kernel machine surrogate models correlate with the underlying neural network. We introduce two new random projection variants of approximate eNTK which allow users to tune the time and memory complexity of their calculation. We conclude that kernel machines using approximate neural tangent kernel as the kernel function are effective surrogate models, with the introduced trace NTK the most consistent performer. Open source software allowing users to efficiently calculate kernel functions in the PyTorch framework is available (https://github.com/pnnl/projection\_ntk).

Explaining RL Decisions with Trajectories

Shripad Vilasrao Deshmukh, Arpan Dasgupta, Balaji Krishnamurthy +4

Explanation is a key component for the adoption of reinforcement learning (RL) in many real-world decision-making problems. In the literature, the explanation is often provided by saliency attribution to the features of the RL agent's state. In this work, we propose a complementary approach to these explanations, particularly for offline RL, where we attribute the policy decisions of a trained RL agent to the trajectories encountered by it during training. To do so, we encode trajectories in offline training data individually as well as collectively (encoding a set of trajectories). We then attribute policy decisions to a set of trajectories in this encoded space by estimating the sensitivity of the decision with respect to that set. Further, we demonstrate the effectiveness of the proposed approach in terms of quality of attributions as well as practical scalability in diverse environments that involve both discrete and continuous state and action spaces such as grid-worlds, video games (Atari) and continuous control (MuJoCo). We also conduct a human study on a simple navigation task to observe how their understanding of the task compares with data attributed for a trained RL policy. Keywords -- Explainable AI, Verifiability of AI Decisions, Explainable RL.

Explaining the behavior of AI systems is an important problem that, in practice, is generally avoided. While the XAI community has been developing an abundance of techniques, most incur a set of costs that the wider deep learning community has been unwilling to pay in most situations. We take a pragmatic view of the issue, and define a set of desiderata that capture both the ambitions of XAI and the practical constraints of deep learning. We describe an effective way to satisfy all the desiderata: train the AI system to build a causal model of itself. We develop an instance of this solution for Deep RL agents: Causal Self-Talk. CST operates by training the agent to communicate with itself across time. We implement this method in a simulated 3D environment, and show how it enables agents to generate faithful and semantically-meaningful explanations of their own behavior. Beyond explanations, we also demonstrate that these learned models provide new ways of building semantic control interfaces to AI systems.

Towards Human-Compatible XAI: Explaining Data Differentials with Concept Induction over Background Knowledge

Cara Widmer, Md Kamruzzaman Sarker, Srikanth Nadella +6

Concept induction, which is based on formal logical reasoning over description logics, has been used in ontology engineering in order to create ontology (TBox) axioms from the base data (ABox) graph. In this paper, we show that it can also be used to explain data differentials, for example in the context of Explainable AI (XAI), and we show that it can in fact be done in a way that is meaningful to a human observer. Our approach utilizes a large class hierarchy, curated from the Wikipedia category hierarchy, as background knowledge.

Neural Payoff Machines: Predicting Fair and Stable Payoff Allocations Among Team Members

Daphne Cornelisse, Thomas Rood, Mateusz Malinowski +2

In many multi-agent settings, participants can form teams to achieve collective outcomes that may far surpass their individual capabilities. Measuring the relative contributions of agents and allocating them shares of the reward that promote long-lasting cooperation are difficult tasks. Cooperative game theory offers solution concepts identifying distribution schemes, such as the Shapley value, that fairly reflect the contribution of individuals to the performance of the team or the Core, which reduces the incentive of agents to abandon their team. Applications of such methods include identifying influential features and sharing the costs of joint ventures or team formation. Unfortunately, using these solutions requires tackling a computational barrier as they are hard to compute, even in restricted settings. In this work, we show how cooperative game-theoretic solutions can be distilled into a learned model by training neural networks to propose fair and stable payoff allocations. We show that our approach creates models that can generalize to games far from the training distribution and can predict solutions for more players than observed during training. An important application of our framework is Explainable AI: our approach can be used to speed-up Shapley value computations on many instances.

Existing work on fairness modeling commonly assumes that sensitive attributes for all instances are fully available, which may not be true in many real-world applications due to the high cost of acquiring sensitive information. When sensitive attributes are not disclosed or available, it is needed to manually annotate a small part of the training data to mitigate bias. However, the skewed distribution across different sensitive groups preserves the skewness of the original dataset in the annotated subset, which leads to non-optimal bias mitigation. To tackle this challenge, we propose Active Penalization Of Discrimination (APOD), an interactive framework to guide the limited annotations towards maximally eliminating the effect of algorithmic bias. The proposed APOD integrates discrimination penalization with active instance selection to efficiently utilize the limited annotation budget, and it is theoretically proved to be capable of bounding the algorithmic bias. According to the evaluation on five benchmark datasets, APOD outperforms the state-of-the-arts baseline methods under the limited annotation budget, and shows comparable performance to fully annotated bias mitigation, which demonstrates that APOD could benefit real-world applications when sensitive information is limited.

SESS: Saliency Enhancing with Scaling and Sliding

Osman Tursun, Simon Denman, Sridha Sridharan +1

High-quality saliency maps are essential in several machine learning application areas including explainable AI and weakly supervised object detection and segmentation. Many techniques have been developed to generate better saliency using neural networks. However, they are often limited to specific saliency visualisation methods or saliency issues. We propose a novel saliency enhancing approach called SESS (Saliency Enhancing with Scaling and Sliding). It is a method and model agnostic extension to existing saliency map generation methods. With SESS, existing saliency approaches become robust to scale variance, multiple occurrences of target objects, presence of distractors and generate less noisy and more discriminative saliency maps. SESS improves saliency by fusing saliency maps extracted from multiple patches at different scales from different areas, and combines these individual maps using a novel fusion scheme that incorporates channel-wise weights and spatial weighted average. To improve efficiency, we introduce a pre-filtering step that can exclude uninformative saliency maps to improve efficiency while still enhancing overall results. We evaluate SESS on object recognition and detection benchmarks where it achieves significant improvement. The code is released publicly to enable researchers to verify performance and further development. Code is available at: https://github.com/neouyghur/SESS

A popular explainable AI (XAI) approach to quantify feature importance of a given model is via Shapley values. These Shapley values arose in cooperative games, and hence a critical ingredient to compute these in an XAI context is a so-called value function, that computes the "value" of a subset of features, and which connects machine learning models to cooperative games. There are many possible choices for such value functions, which broadly fall into two categories: on-manifold and off-manifold value functions, which take an observational and an interventional viewpoint respectively. Both these classes however have their respective flaws, where on-manifold value functions violate key axiomatic properties and are computationally expensive, while off-manifold value functions pay less heed to the data manifold and evaluate the model on regions for which it wasn't trained. Thus, there is no consensus on which class of value functions to use. In this paper, we show that in addition to these existing issues, both classes of value functions are prone to adversarial manipulations on low density regions. We formalize the desiderata of value functions that respect both the model and the data manifold in a set of axioms and are robust to perturbation on off-manifold regions, and show that there exists a unique value function that satisfies these axioms, which we term the Joint Baseline value function, and the resulting Shapley value the Joint Baseline Shapley (JBshap), and validate the effectiveness of JBshap in experiments.