Various fixes

_posts/2024-04-18-a-hanna24a.md

@@ -0,0 +1,57 @@
+---
+title: Multi-Agent Bandit Learning through Heterogeneous Action Erasure Channels
+software: https://github.com/mervekarakas/mamab_erasures
+abstract: Multi-Armed Bandit (MAB) systems are witnessing an upswing in applications
+  within multi-agent distributed environments, leading to the advancement of collaborative
+  MAB algorithms. In such settings, communication between agents executing actions
+  and the primary learner making decisions can hinder the learning process. A prevalent
+  challenge in distributed learning is action erasure, often induced by communication
+  delays and/or channel noise. This results in agents possibly not receiving the intended
+  action from the learner, subsequently leading to misguided feedback. In this paper,
+  we introduce novel algorithms that enable learners to interact concurrently with
+  distributed agents across heterogeneous action erasure channels with different action
+  erasure probabilities. We illustrate that, in contrast to existing bandit algorithms,
+  which experience linear regret, our algorithms assure sub-linear regret guarantees.
+  Our proposed solutions are founded on a meticulously crafted repetition protocol
+  and scheduling of learning across heterogeneous channels. To our knowledge, these
+  are the first algorithms capable of effectively learning through heterogeneous action
+  erasure channels. We substantiate the superior performance of our algorithm through
+  numerical experiments, emphasizing their practical significance in addressing issues
+  related to communication constraints and delays in multi-agent environments.
+layout: inproceedings
+series: Proceedings of Machine Learning Research
+publisher: PMLR
+issn: 2640-3498
+id: a-hanna24a
+month: 0
+tex_title: Multi-Agent Bandit Learning through Heterogeneous Action Erasure Channels
+firstpage: 3898
+lastpage: 3906
+page: 3898-3906
+order: 3898
+cycles: false
+bibtex_author: A Hanna, Osama and Karakas, Merve and Yang, Lin and Fragouli, Christina
+author:
+- given: Osama
+  family: A Hanna
+- given: Merve
+  family: Karakas
+- given: Lin
+  family: Yang
+- given: Christina
+  family: Fragouli
+date: 2024-04-18
+address:
+container-title: Proceedings of The 27th International Conference on Artificial Intelligence
+  and Statistics
+volume: '238'
+genre: inproceedings
+issued:
+  date-parts:
+  - 2024
+  - 4
+  - 18
+pdf: https://proceedings.mlr.press/v238/a-hanna24a/a-hanna24a.pdf
+extras: []
+# Format based on Martin Fenner's citeproc: https://blog.front-matter.io/posts/citeproc-yaml-for-bibliographies/
+---

_posts/2024-04-18-abbas24a.md

@@ -1,7 +1,7 @@
 ---
-title: " Enhancing In-context Learning via Linear Probe Calibration "
-software: " https://github.com/mominabbass/LinC "
-abstract: " In-context learning (ICL) is a new paradigm for natural language processing
+title: Enhancing In-context Learning via Linear Probe Calibration
+software: https://github.com/mominabbass/LinC
+abstract: In-context learning (ICL) is a new paradigm for natural language processing
   that utilizes Generative Pre-trained Transformer (GPT)-like models. This approach
   uses prompts that include in-context demonstrations to generate the corresponding
   output for a new query input. However, applying ICL in real cases does not scale
@@ -16,14 +16,14 @@ abstract: " In-context learning (ICL) is a new paradigm for natural language pro
   of up to 21%, and up to a 50% improvement in some cases, and significantly boosts
   the performance of PEFT methods, especially in the low resource regime. Moreover,
   LinC achieves lower expected calibration error, and is highly robust to varying
-  label proportions, prompt templates, and demonstration permutations. "
+  label proportions, prompt templates, and demonstration permutations.
 layout: inproceedings
 series: Proceedings of Machine Learning Research
 publisher: PMLR
 issn: 2640-3498
 id: abbas24a
 month: 0
-tex_title: " Enhancing In-context Learning via Linear Probe Calibration "
+tex_title: Enhancing In-context Learning via Linear Probe Calibration
 firstpage: 307
 lastpage: 315
 page: 307-315

_posts/2024-04-18-abedsoltan24a.md

@@ -1,6 +1,6 @@
 ---
-title: " On the Nyström Approximation for Preconditioning in Kernel Machines "
-abstract: " Kernel methods are a popular class of nonlinear predictive models in machine
+title: On the Nyström Approximation for Preconditioning in Kernel Machines
+abstract: Kernel methods are a popular class of nonlinear predictive models in machine
   learning. Scalable algorithms for learning kernel models need to be iterative in
   nature, but convergence can be slow due to poor conditioning. Spectral preconditioning
   is an important tool to speed-up the convergence of such iterative algorithms for
@@ -12,14 +12,14 @@ abstract: " Kernel methods are a popular class of nonlinear predictive models in
   such an approximated preconditioner. Specifically, we show that a sample of logarithmic
   size (as a function of the size of the dataset) enables the Nyström-based approximated
   preconditioner to accelerate gradient descent nearly as well as the exact preconditioner,
-  while also reducing the computational and storage overheads. "
+  while also reducing the computational and storage overheads.
 layout: inproceedings
 series: Proceedings of Machine Learning Research
 publisher: PMLR
 issn: 2640-3498
 id: abedsoltan24a
 month: 0
-tex_title: " On the {N}yström Approximation for Preconditioning in Kernel Machines "
+tex_title: On the {N}yström Approximation for Preconditioning in Kernel Machines
 firstpage: 3718
 lastpage: 3726
 page: 3718-3726

_posts/2024-04-18-abernethy24a.md

@@ -0,0 +1,58 @@
+---
+title: 'Lexicographic Optimization: Algorithms and Stability'
+abstract: 'A lexicographic maximum of a set $X \subseteq R^n$ is a vector in $X$ whose
+  smallest component is as large as possible, and subject to that requirement, whose
+  second smallest component is as large as possible, and so on for the third smallest
+  component, etc. Lexicographic maximization has numerous practical and theoretical
+  applications, including fair resource allocation, analyzing the implicit regularization
+  of learning algorithms, and characterizing refinements of game-theoretic equilibria.
+  We prove that a minimizer in $X$ of the exponential loss function $L_c(x) = \sum_i
+  \exp(-c x_i)$ converges to a lexicographic maximum of $X$ as $c \to \infty$, provided
+  that $X$ is \emph{stable} in the sense that a well-known iterative method for finding
+  a lexicographic maximum of $X$ cannot be made to fail simply by reducing the required
+  quality of each iterate by an arbitrarily tiny degree. Our result holds for both
+  near and exact minimizers of the exponential loss, while earlier convergence results
+  made much stronger assumptions about the set $X$ and only held for the exact minimizer.
+  We are aware of no previous results showing a connection between the iterative method
+  for computing a lexicographic maximum and exponential loss minimization. We show
+  that every convex polytope is stable, but that there exist compact, convex sets
+  that are not stable. We also provide the first analysis of the convergence rate
+  of an exponential loss minimizer (near or exact) and discover a curious dichotomy:
+  While the two smallest components of the vector converge to the lexicographically
+  maximum values very quickly (at roughly the rate $\frac{\log n}{c}$), all other
+  components can converge arbitrarily slowly.'
+layout: inproceedings
+series: Proceedings of Machine Learning Research
+publisher: PMLR
+issn: 2640-3498
+id: abernethy24a
+month: 0
+tex_title: 'Lexicographic Optimization: Algorithms and Stability'
+firstpage: 2503
+lastpage: 2511
+page: 2503-2511
+order: 2503
+cycles: false
+bibtex_author: Abernethy, Jacob A. and Schapire, Robert and Syed, Umar
+author:
+- given: Jacob A.
+  family: Abernethy
+- given: Robert
+  family: Schapire
+- given: Umar
+  family: Syed
+date: 2024-04-18
+address:
+container-title: Proceedings of The 27th International Conference on Artificial Intelligence
+  and Statistics
+volume: '238'
+genre: inproceedings
+issued:
+  date-parts:
+  - 2024
+  - 4
+  - 18
+pdf: https://proceedings.mlr.press/v238/abernethy24a/abernethy24a.pdf
+extras: []
+# Format based on Martin Fenner's citeproc: https://blog.front-matter.io/posts/citeproc-yaml-for-bibliographies/
+---

_posts/2024-04-18-abroshan24a.md

@@ -1,6 +1,6 @@
 ---
-title: " Imposing Fairness Constraints in Synthetic Data Generation "
-abstract: " In several real-world applications (e.g., online advertising, item recommendations,
+title: Imposing Fairness Constraints in Synthetic Data Generation
+abstract: In several real-world applications (e.g., online advertising, item recommendations,
   etc.) it may not be possible to release and share the real dataset due to privacy
   concerns. As a result, synthetic data generation (SDG) has emerged as a promising
   solution for data sharing. While the main goal of private SDG is to create a dataset
@@ -12,14 +12,14 @@ abstract: " In several real-world applications (e.g., online advertising, item r
   definition of fairness in synthetic data generation and provide a general framework
   to achieve fairness. Then we consider two notions of counterfactual fairness and
   information filtering fairness and show how our framework can be used for these
-  definitions. "
+  definitions.
 layout: inproceedings
 series: Proceedings of Machine Learning Research
 publisher: PMLR
 issn: 2640-3498
 id: abroshan24a
 month: 0
-tex_title: " Imposing Fairness Constraints in Synthetic Data Generation "
+tex_title: Imposing Fairness Constraints in Synthetic Data Generation
 firstpage: 2269
 lastpage: 2277
 page: 2269-2277

_posts/2024-04-18-achddou24a.md

@@ -1,22 +1,22 @@
 ---
-title: " Multitask Online Learning: Listen to the Neighborhood Buzz "
-abstract: " We study multitask online learning in a setting where agents can only
-  exchange information with their neighbors on an arbitrary communication network.
-  We introduce MT-CO\\textsubscript{2}OL, a decentralized algorithm for this setting
-  whose regret depends on the interplay between the task similarities and the network
-  structure. Our analysis shows that the regret of MT-CO\\textsubscript{2}OL is never
-  worse (up to constants) than the bound obtained when agents do not share information.
-  On the other hand, our bounds significantly improve when neighboring agents operate
-  on similar tasks. In addition, we prove that our algorithm can be made differentially
+title: 'Multitask Online Learning: Listen to the Neighborhood Buzz'
+abstract: We study multitask online learning in a setting where agents can only exchange
+  information with their neighbors on an arbitrary communication network. We introduce
+  MT-CO\textsubscript{2}OL, a decentralized algorithm for this setting whose regret
+  depends on the interplay between the task similarities and the network structure.
+  Our analysis shows that the regret of MT-CO\textsubscript{2}OL is never worse (up
+  to constants) than the bound obtained when agents do not share information. On the
+  other hand, our bounds significantly improve when neighboring agents operate on
+  similar tasks. In addition, we prove that our algorithm can be made differentially
   private with a negligible impact on the regret. Finally, we provide experimental
-  support for our theory. "
+  support for our theory.
 layout: inproceedings
 series: Proceedings of Machine Learning Research
 publisher: PMLR
 issn: 2640-3498
 id: achddou24a
 month: 0
-tex_title: " Multitask Online Learning: Listen to the Neighborhood Buzz "
+tex_title: 'Multitask Online Learning: Listen to the Neighborhood Buzz'
 firstpage: 1846
 lastpage: 1854
 page: 1846-1854

_posts/2024-04-18-adachi24a.md

@@ -1,7 +1,7 @@
 ---
-title: " Looping in the Human: Collaborative and Explainable Bayesian Optimization "
-software: " https://github.com/ma921/CoExBO "
-abstract: " Like many optimizers, Bayesian optimization often falls short of gaining
+title: 'Looping in the Human: Collaborative and Explainable Bayesian Optimization'
+software: https://github.com/ma921/CoExBO
+abstract: Like many optimizers, Bayesian optimization often falls short of gaining
   user trust due to opacity. While attempts have been made to develop human-centric
   optimizers, they typically assume user knowledge is well-specified and error-free,
   employing users mainly as supervisors of the optimization process. We relax these
@@ -15,14 +15,14 @@ abstract: " Like many optimizers, Bayesian optimization often falls short of gai
   even with extreme adversarial interventions, the algorithm converges asymptotically
   to a vanilla Bayesian optimization. We validate CoExBO’s efficacy through human-AI
   teaming experiments in lithium-ion battery design, highlighting substantial improvements
-  over conventional methods. Code is available https://github.com/ma921/CoExBO. "
+  over conventional methods. Code is available https://github.com/ma921/CoExBO.
 layout: inproceedings
 series: Proceedings of Machine Learning Research
 publisher: PMLR
 issn: 2640-3498
 id: adachi24a
 month: 0
-tex_title: " Looping in the Human: Collaborative and Explainable {B}ayesian Optimization "
+tex_title: 'Looping in the Human: Collaborative and Explainable {B}ayesian Optimization'
 firstpage: 505
 lastpage: 513
 page: 505-513

_posts/2024-04-18-adachi24b.md

@@ -1,7 +1,7 @@
 ---
-title: " Adaptive Batch Sizes for Active Learning: A Probabilistic Numerics Approach "
-software: " https://github.com/ma921/AdaBatAL "
-abstract: " Active learning parallelization is widely used, but typically relies on
+title: 'Adaptive Batch Sizes for Active Learning: A Probabilistic Numerics Approach'
+software: https://github.com/ma921/AdaBatAL
+abstract: Active learning parallelization is widely used, but typically relies on
   fixing the batch size throughout experimentation. This fixed approach is inefficient
   because of a dynamic trade-off between cost and speed—larger batches are more costly,
   smaller batches lead to slower wall-clock run-times—and the trade-off may change
@@ -16,14 +16,14 @@ abstract: " Active learning parallelization is widely used, but typically relies
   in the precision requirement, to subsequently adapt batch construction. Through
   extensive experiments, we demonstrate that our approach significantly enhances learning
   efficiency and flexibility in diverse Bayesian batch active learning and Bayesian
-  optimization applications. "
+  optimization applications.
 layout: inproceedings
 series: Proceedings of Machine Learning Research
 publisher: PMLR
 issn: 2640-3498
 id: adachi24b
 month: 0
-tex_title: " Adaptive Batch Sizes for Active Learning: A Probabilistic Numerics Approach "
+tex_title: 'Adaptive Batch Sizes for Active Learning: A Probabilistic Numerics Approach'
 firstpage: 496
 lastpage: 504
 page: 496-504

_posts/2024-04-18-adibi24a.md

@@ -1,35 +1,35 @@
 ---
-title: " Stochastic Approximation with Delayed Updates: Finite-Time Rates under Markovian
-  Sampling "
-abstract: " Motivated by applications in large-scale and multi-agent reinforcement
-  learning, we study the non-asymptotic performance of stochastic approximation (SA)
-  schemes with delayed updates under Markovian sampling. While the effect of delays
-  has been extensively studied for optimization, the manner in which they interact
-  with the underlying Markov process to shape the finite-time performance of SA remains
-  poorly understood. In this context, our first main contribution is to show that
-  under time-varying bounded delays, the delayed SA update rule guarantees exponentially
-  fast convergence of the \\emph{last iterate} to a ball around the SA operator’s
-  fixed point. Notably, our bound is \\emph{tight} in its dependence on both the maximum
-  delay $\\tau_{max}$, and the mixing time $\\tau_{mix}$. To achieve this tight bound,
-  we develop a novel inductive proof technique that, unlike various existing delayed-optimization
-  analyses, relies on establishing uniform boundedness of the iterates. As such, our
-  proof may be of independent interest. Next, to mitigate the impact of the maximum
-  delay on the convergence rate, we provide the first finite-time analysis of a delay-adaptive
+title: 'Stochastic Approximation with Delayed Updates: Finite-Time Rates under Markovian
+  Sampling'
+abstract: Motivated by applications in large-scale and multi-agent reinforcement learning,
+  we study the non-asymptotic performance of stochastic approximation (SA) schemes
+  with delayed updates under Markovian sampling. While the effect of delays has been
+  extensively studied for optimization, the manner in which they interact with the
+  underlying Markov process to shape the finite-time performance of SA remains poorly
+  understood. In this context, our first main contribution is to show that under time-varying
+  bounded delays, the delayed SA update rule guarantees exponentially fast convergence
+  of the \emph{last iterate} to a ball around the SA operator’s fixed point. Notably,
+  our bound is \emph{tight} in its dependence on both the maximum delay $\tau_{max}$,
+  and the mixing time $\tau_{mix}$. To achieve this tight bound, we develop a novel
+  inductive proof technique that, unlike various existing delayed-optimization analyses,
+  relies on establishing uniform boundedness of the iterates. As such, our proof may
+  be of independent interest. Next, to mitigate the impact of the maximum delay on
+  the convergence rate, we provide the first finite-time analysis of a delay-adaptive
   SA scheme under Markovian sampling. In particular, we show that the exponent of
-  convergence of this scheme gets scaled down by $\\tau_{avg}$, as opposed to $\\tau_{max}$
-  for the vanilla delayed SA rule; here, $\\tau_{avg}$ denotes the average delay across
+  convergence of this scheme gets scaled down by $\tau_{avg}$, as opposed to $\tau_{max}$
+  for the vanilla delayed SA rule; here, $\tau_{avg}$ denotes the average delay across
   all iterations. Moreover, the adaptive scheme requires no prior knowledge of the
   delay sequence for step-size tuning. Our theoretical findings shed light on the
   finite-time effects of delays for a broad class of algorithms, including TD learning,
-  Q-learning, and stochastic gradient descent under Markovian sampling. "
+  Q-learning, and stochastic gradient descent under Markovian sampling.
 layout: inproceedings
 series: Proceedings of Machine Learning Research
 publisher: PMLR
 issn: 2640-3498
 id: adibi24a
 month: 0
-tex_title: " Stochastic Approximation with Delayed Updates: Finite-Time Rates under
-  {M}arkovian Sampling "
+tex_title: 'Stochastic Approximation with Delayed Updates: Finite-Time Rates under
+  {M}arkovian Sampling'
 firstpage: 2746
 lastpage: 2754
 page: 2746-2754

_posts/2024-04-18-aghbalou24a.md

@@ -1,7 +1,7 @@
 ---
-title: " Sharp error bounds for imbalanced classification: how many examples in the
-  minority class? "
-abstract: " When dealing with imbalanced classification data, reweighting the loss
+title: 'Sharp error bounds for imbalanced classification: how many examples in the
+  minority class?'
+abstract: 'When dealing with imbalanced classification data, reweighting the loss
   function is a standard procedure allowing to equilibrate between the true positive
   and true negative rates within the risk measure. Despite significant theoretical
   work in this area, existing results do not adequately address a main challenge within
@@ -12,15 +12,15 @@ abstract: " When dealing with imbalanced classification data, reweighting the lo
   fast rate probability bound for constrained balanced empirical risk minimization,
   and (2) a consistent upper bound for balanced nearest neighbors estimates. Our findings
   provide a clearer understanding of the benefits of class-weighting in realistic
-  settings, opening new avenues for further research in this field. "
+  settings, opening new avenues for further research in this field.'
 layout: inproceedings
 series: Proceedings of Machine Learning Research
 publisher: PMLR
 issn: 2640-3498
 id: aghbalou24a
 month: 0
-tex_title: " Sharp error bounds for imbalanced classification: how many examples in
-  the minority class? "
+tex_title: 'Sharp error bounds for imbalanced classification: how many examples in
+  the minority class?'
 firstpage: 838
 lastpage: 846
 page: 838-846