Stochastic and adversarial data are two widely studied settings in online learning. But many optimization tasks are neither i.i.d. nor fully adversarial, which makes it of fundamental interest to get a better theoretical understanding of the world between these extremes. In this work we establish novel regret bounds for online convex optimization in a setting that interpolates between stochastic i.i.d. and fully adversarial losses. By exploiting smoothness of the expected losses, these bounds replace a dependence on the maximum gradient length by the variance of the gradients, which was previously known only for linear losses. In addition, they weaken the i.i.d. assumption by allowing, for example, adversarially poisoned rounds, which were previously considered in the related expert and bandit settings. In the fully i.i.d. case, our regret bounds match the rates one would expect from results in stochastic acceleration, and we also recover the optimal stochastically accelerated rates via online-to-batch conversion. In the fully adversarial case our bounds gracefully deteriorate to match the minimax regret. We further provide lower bounds showing that our regret upper bounds are tight for all intermediate regimes in terms of the stochastic variance and the adversarial variation of the loss gradients.

Learning on a massive amount of speech corpus leads to the recent success of many self-supervised speech models. With knowledge distillation, these models may also benefit from the knowledge encoded by language models that are pre-trained on rich sources of texts. The distillation process, however, is challenging due to the modal disparity between textual and speech embedding spaces. This paper studies metric-based distillation to align the embedding space of text and speech with only a small amount of data without modifying the model structure. Since the semantic and granularity gap between text and speech has been omitted in literature, which impairs the distillation, we propose the Prior-informed Adaptive knowledge Distillation (PAD) that adaptively leverages text/speech units of variable granularity and prior distributions to achieve better global and local alignments between text and speech pre-trained models. We evaluate on three spoken language understanding benchmarks to show that PAD is more effective in transferring linguistic knowledge than other metric-based distillation approaches.

Open-ended learning methods that automatically generate a curriculum of increasingly challenging tasks serve as a promising avenue toward generally capable reinforcement learning agents. Existing methods adapt curricula independently over either environment parameters (in single-agent settings) or co-player policies (in multi-agent settings). However, the strengths and weaknesses of co-players can manifest themselves differently depending on environmental features. It is thus crucial to consider the dependency between the environment and co-player when shaping a curriculum in multi-agent domains. In this work, we use this insight and extend Unsupervised Environment Design (UED) to multi-agent environments. We then introduce Multi-Agent Environment Design Strategist for Open-Ended Learning (MAESTRO), the first multi-agent UED approach for two-player zero-sum settings. MAESTRO efficiently produces adversarial, joint curricula over both environments and co-players and attains minimax-regret guarantees at Nash equilibrium. Our experiments show that MAESTRO outperforms a number of strong baselines on competitive two-player games, spanning discrete and continuous control settings.

Deep learning models for learning analytics have become increasingly popular over the last few years; however, these approaches are still not widely adopted in real-world settings, likely due to a lack of trust and transparency. In this paper, we tackle this issue by implementing explainable AI methods for black-box neural networks. This work focuses on the context of online and blended learning and the use case of student success prediction models. We use a pairwise study design, enabling us to investigate controlled differences between pairs of courses. Our analyses cover five course pairs that differ in one educationally relevant aspect and two popular instance-based explainable AI methods (LIME and SHAP). We quantitatively compare the distances between the explanations across courses and methods. We then validate the explanations of LIME and SHAP with 26 semi-structured interviews of university-level educators regarding which features they believe contribute most to student success, which explanations they trust most, and how they could transform these insights into actionable course design decisions. Our results show that quantitatively, explainers significantly disagree with each other about what is important, and qualitatively, experts themselves do not agree on which explanations are most trustworthy. All code, extended results, and the interview protocol are provided at https://github.com/epfl-ml4ed/trusting-explainers.

A data marketplace is an online venue that brings data owners, data brokers, and data consumers together and facilitates commoditisation of data amongst them. Data pricing, as a key function of a data marketplace, demands quantifying the monetary value of data. A considerable number of studies on data pricing can be found in literature. This paper attempts to comprehensively review the state-of-the-art on existing data pricing studies to provide a general understanding of this emerging research area. Our key contribution lies in a new taxonomy of data pricing studies that unifies different attributes determining data prices. The basis of our framework categorises these studies by the kind of market structure, be it sell-side, buy-side, or two-sided. Then in a sell-side market, the studies are further divided by query type, which defines the way a data consumer accesses data, while in a buy-side market, the studies are divided according to privacy notion, which defines the way to quantify privacy of data owners. In a two-sided market, both privacy notion and query type are used as criteria. We systematically examine the studies falling into each category in our taxonomy. Lastly, we discuss gaps within the existing research and define future research directions.

Traditional models of opinion dynamics provide a simple approach to understanding human behavior in basic social scenarios. However, when it comes to issues such as polarization and extremism, we require a more nuanced understanding of human biases and cognitive tendencies. In this paper, we propose an approach to modeling opinion dynamics by integrating mental models and assumptions of individuals agents using Bayesian-inspired methods. By exploring the relationship between human rationality and Bayesian theory, we demonstrate the efficacy of these methods in describing how opinions evolve. Our analysis leverages the Continuous Opinions and Discrete Actions (CODA) model, applying Bayesian-inspired rules to account for key human behaviors such as confirmation bias, motivated reasoning, and our reluctance to change opinions. Through this, we obtain update rules that offer deeper insights into the dynamics of extreme opinions. Our work sheds light on the role of human biases in shaping opinion dynamics and highlights the potential of Bayesian-inspired modeling to provide more accurate predictions of real-world scenarios. Keywords: Opinion dynamics, Bayesian methods, Cognition, CODA, Agent-based models

Training deep neural networks is a challenging non-convex optimization problem. Recent work has proven that the strong duality holds (which means zero duality gap) for regularized finite-width two-layer ReLU networks and consequently provided an equivalent convex training problem. However, extending this result to deeper networks remains to be an open problem. In this paper, we prove that the duality gap for deeper linear networks with vector outputs is non-zero. In contrast, we show that the zero duality gap can be obtained by stacking standard deep networks in parallel, which we call a parallel architecture, and modifying the regularization. Therefore, we prove the strong duality and existence of equivalent convex problems that enable globally optimal training of deep networks. As a by-product of our analysis, we demonstrate that the weight decay regularization on the network parameters explicitly encourages low-rank solutions via closed-form expressions. In addition, we show that strong duality holds for three-layer standard ReLU networks given rank-1 data matrices.

This paper presents a novel study of parameter-free attentive scoring for speaker verification. Parameter-free scoring provides the flexibility of comparing speaker representations without the need of an accompanying parametric scoring model. Inspired by the attention component in Transformer neural networks, we propose a variant of the scaled dot product attention mechanism to compare enrollment and test segment representations. In addition, this work explores the effect on performance of (i) different types of normalization, (ii) independent versus tied query/key estimation, (iii) varying the number of key-value pairs and (iv) pooling multiple enrollment utterance statistics. Experimental results for a 4 task average show that a simple parameter-free attentive scoring mechanism can improve the average EER by 10% over the best cosine similarity baseline.

There is too much internet and our attempts to keep up with the breakneck pace of, well, everything these days -- it is breaking our brains. Parsing through the deluge of inundating information hoisted up by algorithmic systems built to maximize engagement has trained us as slavering Pavlovian dogs to rely on snap judgements and gut feelings in our decision making and opinion formation rather than deliberation and introspection. Which is fine when you're deciding between Italian and Indian for dinner or are waffling on a new paint color for the hallway, but not when we're out here basing existential life choices on friggin' vibes. In his latest book, I, HUMAN: AI, Automation, and the Quest to Reclaim What Makes Us Unique, professor of business psychology and Chief Innovation Officer at ManpowerGroup, Tomas Chamorro-Premuzic explores the myriad ways that AI systems now govern our daily lives and interactions. From finding love to finding gainful employment to finding out the score of yesterday's game, AI has streamlined the information gathering process.

Conformal prediction [1, 2, 3, 4, 5, 6], a technique developed to address the confidence in the forecasts made by general predictive models, is quickly moving the field of machine learning [7, 8, 9, 10] from a period dominated by point predictions, to a new stage in which inferences about the future are summarized by prediction sets with statistical guarantees. Several features make conformal prediction appealing for use with contemporary machine learning algorithms: it is universal (distribution-free), able to handle high-dimensional data, model agnostic, and its properties hold for finite samples. This paper strengthens the universal properties of the most readily applicable variation on the conformal prediction idea: the split conformal prediction algorithm [2, 5], whose implementation attains a good balance between the predictive goals and the computational complexity of the procedure. In a regression context, the main results of the paper are the identification for exchangeable data of the exact distribution of the coverage of prediction sets for a finite horizon of future observables (future coverage, for short), and the determination of the exact distribution of its almost sure limit when the horizon tends to infinity.