United States

When inferring reward functions from human behavior (be it demonstrations, comparisons, physical corrections, or e-stops), it has proven useful to model the human as making noisy-rational choices, with a ``rationality coefficient&quot; capturing how much noise or entropy we expect to see in the human behavior. Prior work typically sets the rationality level to a constant value, regardless of the type, or quality, of human feedback. However, in many settings, giving one type of feedback (e.g. a demonstration) may be much more difficult than a different type of feedback (e.g. answering a comparison query). Thus, we expect to see more or less noise or suboptimality depending on the type of human feedback. In this work, we advocate that grounding the rationality coefficient in real data for each feedback type, rather than assuming a default value, has a significant positive effect on reward learning. We test this in both simulated experiments and in a user study with real human feedback. We find that overestimating human rationality can have dire effects on reward learning accuracy and regret. We also find that fitting the rationality coefficient to human data enables better reward learning, even when the human deviates significantly from the noisy-rational choice model due to systematic biases. Further, we find that the rationality level affects the informativeness of each feedback type: surprisingly, demonstrations are not always the most informative---when the human acts very suboptimally, comparisons actually become more informative, even when the rationality level is the same for both.  Ultimately, our results emphasize the importance and advantage of paying attention to the assumed human-rationality-level, especially when agents actively learn from multiple types of human feedback.

AAAI 2023

The Effect of Modeling Human Rationality Level on Learning Rewards from Multiple Feedback Types

imitation learning & inverse reinforcement learning

learning human values and preferences

human-in-the-loop machine learning

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.

When inferring reward functions from human behavior (be it demonstrations, comparisons, physical corrections, or e-stops), it has proven useful to model the human as making noisy-rational choices, with a ``rationality coefficient" capturing how much noise or entropy we expect to see in the human behavior. Prior work typically sets the rationality level to a constant value, regardless of the type, or quality, of human feedback. However, in many settings, giving one type of feedback (e.g. a demonstration) may be much more difficult than a different type of feedback (e.g. answering a comparison query). Thus, we expect to see more or less noise or suboptimality depending on the type of human feedback. In this work, we advocate that grounding the rationality coefficient in real data for each feedback type, rather than assuming a default value, has a significant positive effect on reward learning. We test this in both simulated experiments and in a user study with real human feedback. We find that overestimating human rationality can have dire effects on reward learning accuracy and regret. We also find that fitting the rationality coefficient to human data enables better reward learning, even when the human deviates significantly from the noisy-rational choice model due to systematic biases. Further, we find that the rationality level affects the informativeness of each feedback type: surprisingly, demonstrations are not always the most informative---when the human acts very suboptimally, comparisons actually become more informative, even when the rationality level is the same for both.  Ultimately, our results emphasize the importance and advantage of paying attention to the assumed human-rationality-level, especially when agents actively learn from multiple types of human feedback.

poster

Secure aggregation is a critical component in federated learning (FL), which enables the server to learn the aggregate model of the users without observing their local models. Conventionally, secure aggregation algorithms focus only on ensuring the privacy of individual users in a single training round. We contend that such designs can lead to significant privacy leakages over multiple training rounds, due to partial user selection/participation at each round of FL. In fact, we show that the conventional random user selection strategies in FL lead to leaking users' individual models within number of rounds that is linear in the number of users.
To address this challenge, we introduce a secure aggregation framework, Multi-RoundSecAgg, with multi-round privacy guarantees.
In particular, we introduce a new metric to quantify the privacy guarantees of FL over multiple training rounds, and develop a structured user selection strategy that guarantees the long-term privacy of each user (over any number of training rounds). 
Our framework also carefully accounts for the fairness and the average number of participating users at each round.
Our experiments on MNIST and CIFAR-10 datasets in the IID and the non-IID settings demonstrate the performance improvement over the baselines, both in terms of privacy protection and test accuracy. 

Securing Secure Aggregation: Mitigating Multi-Round Privacy Leakage in Federated Learning

Target Propagation (TP) is a biologically more plausible algorithm than the error backpropagation (BP) to train deep networks, and improving practicality of TP is an open issue. TP methods  require the feedforward and feedback networks to form layer-wise autoencoders for propagating the target values generated at the output layer. However, this causes certain drawbacks; e.g., careful hyperparameter tuning is required to synchronize the feedforward and feedback training, and frequent updates of the feedback path are usually required than that of the feedforward path. Learning of the feedforward and feedback networks is sufficient to make TP methods capable of training, but is having these layer-wise autoencoders a necessary condition for TP to work? We answer this question by presenting Fixed-Weight Difference Target Propagation (FW-DTP) that keeps the feedback weights constant during training. We confirmed that this simple method, which naturally resolves the abovementioned problems of TP, can still deliver informative target values to hidden layers for a given task; indeed, FW-DTP consistently achieves higher test performance than a baseline, the Difference Target Propagation (DTP), on four classification datasets. We also present a novel propagation architecture that explains the exact form of the feedback function of DTP to analyze FW-DTP.

Fixed-Weight Difference Target Propagation

Given a square matrix with noisy dissimilarity measures between pairs of data samples, the metric nearness model computes the best approximation of the matrix from a set of valid distance metrics. Despite its wide applications in machine learning and data processing tasks, the model faces non-trivial computational requirements in seeking the solution due to the large number of metric constraints associated with the feasible region. To tackle the challenge and improve the model's scalability and applicability, our work designed a practical approach that operates in two stages. The first stage computes a fast yet high-quality approximate solution from a set of isometrically embeddable metrics, which is further improved by an effective heuristic. The second stage refines the approximate solution with the Halpern-Lions-Wittmann-Bauschke projection algorithm, which converges quickly to the optimal solution. In empirical evaluations, the proposed approach runs at least an order of magnitude faster than the state-of-the-art solutions, with significantly improved scalability, complete conformity to constraints, less memory consumption, and other desirable features in real applications.

Metric Nearness Made Practical

 We present a novel method for exemplar-based image translation, called matching interleaved diffusion models (MIDMs). Most existing methods for this task were formulated as GAN-based matching-then-generation framework. However, in this framework, matching errors induced by the difficulty of semantic matching across cross-domain, e.g., sketch and photo, can be easily propagated to the generation step, which in turn leads to the degenerated results. Motivated by the recent success of diffusion models, overcoming the shortcomings of GANs, we incorporate the diffusion models to overcome these limitations. Specifically, we formulate a diffusion-based matching-and-generation framework that interleaves cross-domain matching and diffusion steps in the latent space by iteratively feeding the intermediate warp into the noising process and denoising it to generate a translated image. In addition, to improve the reliability of diffusion process, we design confidence-aware process using cycle-consistency to consider only confident regions during translation. Experimental results show that our MIDMs generate more plausible images than state-of-the-art methods.

MIDMs: Matching Interleaved Diffusion Models for Exemplar-based Image Translation

The marginal distribution shift of covariates from training to test phase, named covariate-shift, is prone to make a predictive model with model misspecification perform unstably across agnostic testing data. The recent literature on invariant learning attempts to learn an invariant predictor from heterogeneous environments. However, the performance of the learned predictor depends heavily on the availability and quality of provided environments. In this paper, we propose a simple and effective non-parametric method for generating heterogeneous environments via Random Sample Weighting (RSW). Given the training dataset from a single source environment, we randomly generate a set of covariate-determining sample weights and use each weighted training distribution to simulate an environment. We theoretically show that under appropriate conditions, such random sample weighting can produce sufficient heterogeneity to be exploited by common invariance constraints to find invariant variables for stable prediction under covariate shifts. Extensive experiments on both simulated and real-world datasets clearly validate the effectiveness of our method.

Covariate-Shift Generalization via Random Sample Weighting

Transformer-based autoregressive (AR) methods have achieved appealing performance for varied sequence-to-sequence generation tasks, e.g., neural machine translation, summarization, and code generation, but suffer from low inference efficiency. To speed up the inference stage, many non-autoregressive (NAR) strategies have been proposed in the past few years. Among them, the conditional masked language model (CMLM) is one of the most versatile frameworks, as it can support many different sequence generation scenarios and achieve very competitive performance on these tasks. In this paper, we further introduce a simple yet effective adaptive masking over masking strategy to enhance the refinement capability of the decoder and make the encoder optimization easier. Experiments on \textbf{3} different tasks (neural machine translation, summarization, and code generation) with \textbf{15} datasets in total confirm that our proposed simple method achieves significant performance improvement over the strong CMLM model. Surprisingly, our proposed model yields state-of-the-art performance on neural machine translation (\textbf{34.62} BLEU on WMT16 EN$\to$RO, \textbf{34.82} BLEU on WMT16 RO$\to$EN, and \textbf{34.84} BLEU on IWSLT De$\to$En) and even better performance than the \textbf{AR} Transformer on \textbf{7} benchmark datasets with at least \textbf{2.2$\times$} speedup. Our anonymous code is available at GitHub\footnote{\url{ https://github.com/amom-nar/AMOM}}.

AMOM: Adaptive Masking over Masking for Conditional Masked Language Model

Existing knowledge graph (KG) embedding models have primarily focused on static KGs. However, real-world KGs do not remain static, but rather evolve and grow in tandem with the development of KG applications. Consequently, new facts and previously unseen entities and relations continually emerge, necessitating an embedding model that can quickly learn and transfer new knowledge through growth. Motivated by this, we delve into an expanding field of KG embedding in this paper, i.e., lifelong KG embedding. We consider knowledge transfer and retention of the learning on growing snapshots of a KG without having to learn embeddings from scratch. The proposed model includes a masked KG autoencoder for embedding learning and update, with an embedding transfer strategy to inject the learned knowledge into the new entity and relation embeddings, and an embedding regularization method to avoid catastrophic forgetting. To investigate the impacts of different aspects of KG growth, we construct four datasets to evaluate the performance of lifelong KG embedding. Experimental results show that the proposed model outperforms the state-of-the-art inductive and lifelong embedding baselines.

Lifelong Embedding Learning and Transfer for Growing Knowledge Graphs

The prevalence of region-based urban data has enabled us to explore the correlations of regions for better urban planning and smart-city solutions, in which region embedding plays an important role. However, this is a challenging problem considering the following two factors: different city data for region embedding and the effectiveness for various downstream applications. 
In this paper, we propose a novel framework, named \textbf{HREP} (\textbf{H}eterogeneous \textbf{R}egion \textbf{E}mbedding with \textbf{P}rompt learning), considering both intra-region and inter-region correlations, which includes two major modules, \textbf{H}eterogeneous graph learning for \textbf{R}egion \textbf{E}mbedding (\textbf{HRE}) and prompt learning for different downstream tasks. In the HRE module, we construct a region heterogeneous graph according to three categories of data, capturing inter-region contexts, e.g., human mobility and region geographic neighbors, and intra-region contexts, e.g., POI information. Specially, we design a \textit{relation-aware graph embedding} to learn region embedding and relation embeddings of edge types based on the heterogeneous graph. Then we introduce self-attention to capture the global correlation of regions. Moreover, we develop a \textit{attention-based fusion module} to integrate sharing information among different types of correlations. 
To enable region embedding to be effectively utilized in downstream tasks, we migrate the prompt learning, i.e., continuous prompt
prefix-tuning, which can guide the learning of downstream tasks, leading to better prediction results. Extensive experimental results on real-world datasets show that our model outperforms all state-of-the-art methods dramatically.

Heterogeneous Region Embedding with Prompt Learning

In many applications we want to influence the decisions of independent agents by designing incentives for their actions. We revisit a fundamental problem in this area, called Game Implementation: Given a game in standard form and a set of desired strategies, can we design a set of payment promises such that, if the players take the payment promises into account, then all undominated strategies are desired? Furthermore, we aim to minimize the cost, that is, the worst-case amount of payments.

We study the tractability of computing such payment promises and determine more closely what obstructions we may have to overcome in doing so. We show that Game Implementation is NP-hard even for two players, solving in particular a long open question (Eidenbenz et al. 2011). This suggests more restrictions are necessary to obtain tractability. We thus study the regime in which players have only a small constant number of strategies and obtain the following. First, this case remains NP-hard even if each player's utility depends only on three others. Second, we repair a flawed efficient algorithm for the case of both small number of strategies and small number of players. Among further results, we characterize sets of desired strategies that can be implemented at zero cost as a kind of stable core of the game.

Game Implementation: What are the Obstructions?

Digital advertising constitutes one of the main revenue sources for online platforms. In recent years, some advertisers tend to adopt auto-bidding tools to facilitate advertising performance optimization, making the classical \emph{utility maximizer} model in auction theory not fit well. Some recent studies proposed a new model, called \emph{value maximizer}, for auto-bidding advertisers with return-on-investment (ROI) constraints. However, the model of either utility maximizer or value maximizer could only characterize partial advertisers in real-world advertising platforms. In a mixed environment where utility maximizers and value maximizers coexist, the truthful ad auction design would be challenging since bidders could manipulate both their values and affiliated classes, leading to a multi-parameter mechanism design problem. In this work, we address this issue by proposing a payment rule which combines the corresponding ones in classical VCG and GSP mechanisms in a novel way. Based on this payment rule, we propose a truthful auction mechanism with an approximation ratio of $2$ on social welfare, which is close to the lower bound of at least $\frac{5}{4}$ that we also prove. The designed auction mechanism is a generalization of VCG for utility maximizers and GSP for value maximizers.

The Effect of Modeling Human Rationality Level on Learning Rewards from Multiple Feedback Types

Downloads

Next from AAAI 2023

Securing Secure Aggregation: Mitigating Multi-Round Privacy Leakage in Federated Learning

Similar lecture

Human-in-the-Loop Vehicle ReID

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES