ICML 2023 Papers

We study first-order methods with preconditioning for solving structured convex optimization problems. We propose a new family of preconditioners generated by the symmetric polynomials. They provide the first-order optimization methods with a provable improvement of the condition number, cutting the gaps between highest eigenvalues, without explicit knowledge of the actual spectrum. We give a stochastic interpretation of this preconditioning in terms of the coordinate volume sampling and compare it with other classical approaches, including the Chebyshev polynomials. We show how to incorporate a polynomial preconditioning into the Gradient and Fast Gradient Methods and establish their better global complexity bounds. Finally, we propose a simple adaptive search procedure that automatically ensures the best polynomial preconditioning for the Gradient Method, minimizing the objective along a low-dimensional Krylov subspace. Numerical experiments confirm the efficiency of our preconditioning strategies for solving various machine learning problems.

This paper studies the fair range clustering problem in which the data points are from different demographic groups and the goal is to pick $k$ centers with the minimum clustering cost such that each group is at least minimally represented in the centers set and no group dominates the centers set. More precisely, given a set of $n$ points in a metric space $(P, d)$ where each point belongs to one of the $\ell$ different demographics (i.e., $P = P_1 \uplus P_2 \uplus \cdots \uplus P_\ell$) and a set of $\ell$ intervals $[\alpha_1, \beta_1], \cdots, [\alpha_\ell, \beta_\ell]$ on desired number of centers from each group, the goal is to pick a set of $k$ centers $C$ with minimum $\ell_p$-clustering cost (i.e., $(\sum_{v\in P} d(v,C)^p)^{1/p}$) such that for each group $i\in \ell$, $|C\cap P_i| \in [\alpha_i, \beta_i]$. In particular, the fair range $\ell_p$-clustering captures fair range $k$-center, $k$-median and $k$-means as its special cases. In this work, we provide an efficient constant factor approximation algorithm for the fair range $\ell_p$-clustering for all values of $p\in [1,\infty)$.

We present a framework for using transformer networks as universal computers by programming them with specific weights and placing them in a loop. Our input sequence acts as a punchcard, consisting of instructions and memory for data read/writes. We demonstrate that a constant number of encoder layers can emulate basic computing blocks, including lexicographic operations, non-linear functions, function calls, program counters, and conditional branches. Using this framework, we emulate a computer using a simple instruction-set architecture, which allows us to map iterative algorithms to programs that can be executed by a constant depth looped transformer network. We show how a single frozen transformer, instructed by its input, can emulate a basic calculator, a basic linear algebra library, and even a full backpropagation, in-context learning algorithm. Our findings reveal the potential of transformer networks as programmable compute units and offer insight into the mechanics of attention.

In this paper, we study the problem of establishing the accountability and fairness of transparent machine learning models through monotonicity. Although there have been numerous studies on individual monotonicity, pairwise monotonicity is often overlooked in the existing literature. This paper studies transparent neural networks in the presence of three types of monotonicity: individual monotonicity, weak pairwise monotonicity, and strong pairwise monotonicity. As a means of achieving monotonicity while maintaining transparency, we propose the monotonic groves of neural additive models. As a result of empirical examples, we demonstrate that monotonicity is often violated in practice and that monotonic groves of neural additive models are transparent, accountable, and fair.

Chaotic dynamical systems (DS) are ubiquitous in nature and society. Often we are interested in reconstructing such systems from observed time series for prediction or mechanistic insight, where by reconstruction we mean learning geometrical and invariant temporal properties of the system in question (like attractors). However, training reconstruction algorithms like recurrent neural networks (RNNs) on such systems by gradient-descent based techniques faces severe challenges. This is mainly due to exploding gradients caused by the exponential divergence of trajectories in chaotic systems. Moreover, for (scientific) interpretability we wish to have as low dimensional reconstructions as possible, preferably in a model which is mathematically tractable. Here we report that a surprisingly simple modification of teacher forcing leads to provably strictly all-time bounded gradients in training on chaotic systems, and, when paired with a simple architectural rearrangement of a tractable RNN design, piecewise-linear RNNs (PLRNNs), allows for faithful reconstruction in spaces of at most the dimensionality of the observed system. We show on several DS that with these amendments we can reconstruct DS better than current SOTA algorithms, in much lower dimensions. Performance differences were particularly compelling on real world data with which most other methods severely struggled. This work thus led to a simple yet powerful DS reconstruction algorithm which is highly interpretable at the same time.

This paper introduces a distributed, GPU-centric experience replay system, GEAR, designed to perform scalable reinforcement learning (RL) with large sequence models (such as transformers). With such models, existing systems such as Reverb face considerable bottlenecks in memory, computation, and communication. GEAR, however, optimizes memory efficiency by enabling the memory resources on GPU servers (including host memory and device memory) to manage trajectory data. Furthermore, it facilitates decentralized GPU devices to expedite various trajectory selection strategies, circumventing computational bottlenecks. GEAR is equipped with GPU kernels capable of collecting trajectories using zero-copy access to host memory, along with remote-directed-memory access over InfiniBand, improving communication efficiency. Cluster experiments have shown that GEAR can achieve performance levels up to 6× greater than Reverb when training state-of-the-art large RL models. GEAR is open-sourced at https:// github.com/bigrl-team/gear.

Understanding and analyzing markets is crucial, yet analytical equilibrium solutions remain largely infeasible. Recent breakthroughs in equilibrium computation rely on zeroth-order policy gradient estimation. These approaches commonly suffer from high variance and are computationally expensive. The use of fully differentiable simulators would enable more efficient gradient estimation. However, the discrete allocation of goods in economic simulations is a non-differentiable operation. This renders the first-order Monte Carlo gradient estimator inapplicable and the learning feedback systematically misleading. We propose a novel smoothing technique that creates a surrogate market game, in which first-order methods can be applied. We provide theoretical bounds on the resulting bias which justifies solving the smoothed game instead. These bounds also allow choosing the smoothing strength a priori such that the resulting estimate has low variance. Furthermore, we validate our approach via numerous empirical experiments. Our method theoretically and empirically outperforms zeroth-order methods in approximation quality and computational efficiency.

Transformers are central to recent successes in natural language processing and computer vision. Transformers have a mostly uniform backbone where layers alternate between feed-forward and self-attention in order to build a deep network. Here we investigate this design choice and find that more complex blocks that have different permutations of layer primitives can be more efficient. Using this insight, we develop a complex block, named Brainformer, that consists of a diverse sets of layers such as sparsely gated feed-forward layers, dense feed-forward layers, attention layers, and various forms of layer normalization and activation functions. Brainformer consistently outperforms the state-of-the-art dense and sparse Transformers, in terms of both quality and efficiency. A Brainformer model with 8 billion activated parameters per token demonstrates 2x faster training convergence and 5x faster step time compared to its GLaM counterpart. In downstream task evaluation, Brainformer also demonstrates a 3% higher SuperGLUE score with fine-tuning compared to GLaM with a similar number of activated parameters. Finally, Brainformer largely outperforms a Primer dense model derived with NAS with similar computation per token on fewshot evaluations.

The output distribution of a neural network (NN) over the entire input space captures the complete input-output mapping relationship, offering in- sights toward a more comprehensive NN under- standing. Exhaustive enumeration or traditional Monte Carlo methods for the entire input space can exhibit impractical sampling time, especially for high-dimensional inputs. To make such difficult sampling computationally feasible, in this paper, we propose a novel Gradient-based Wang-Landau (GWL) sampler. We first draw the connection between the output distribution of a NN and the density of states (DOS) of a physical system. Then, we renovate the classic sampler for the DOS problem, Wang-Landau algorithm, by re-placing its random proposals with gradient-based Monte Carlo proposals. This way, our GWL sampler investigates the under-explored subsets of the input space much more efficiently. Extensive experiments have verified the accuracy of the output distribution generated by GWL and also showcased several interesting findings - for example, in a binary image classification task, both CNN and ResNet mapped the majority of human unrecognizable images to very negative logit values.

In this paper, we examine the effectiveness of pre-training for visuo-motor control tasks. We revisit a simple Learning-from-Scratch (LfS) baseline that incorporates data augmentation and a shallow ConvNet, and find that this baseline is surprisingly competitive with recent approaches (PVR, MVP, R3M) that leverage frozen visual representations trained on large-scale vision datasets -- across a variety of algorithms, task domains, and metrics in simulation and on a real robot. Our results demonstrate that these methods are hindered by a significant domain gap between the pre-training datasets and current benchmarks for visuo-motor control, which is alleviated by finetuning. Based on our findings, we provide recommendations for future research in pre-training for control and hope that our simple yet strong baseline will aid in accurately benchmarking progress in this area. Code: https://github.com/gemcollector/learning-from-scratch.

We investigate how pair-wise data augmentation techniques like Mixup affect the sample complexity of finding optimal decision boundaries in a binary linear classification problem. For a family of data distributions with a separability constant $\kappa$, we analyze how well the optimal classifier in terms of training loss aligns with the optimal one in test accuracy (i.e., Bayes optimal classifier). For vanilla training without augmentation, we uncover an interesting phenomenon named the curse of separability. As we increase $\kappa$ to make the data distribution more separable, the sample complexity of vanilla training increases exponentially in $\kappa$; perhaps surprisingly, the task of finding optimal decision boundaries becomes harder for more separable distributions. For Mixup training, we show that Mixup mitigates this problem by significantly reducing the sample complexity. To this end, we develop new concentration results applicable to $n^2$ pair-wise augmented data points constructed from $n$ independent data, by carefully dealing with dependencies between overlapping pairs. Lastly, we study other masking-based Mixup-style techniques and show that they can distort the training loss and make its minimizer converge to a suboptimal classifier in terms of test accuracy.

We develop a variant of the stochastic prox-linear method for minimizing the Conditional Value-at-Risk (CVaR) objective. CVaR is a risk measure focused on minimizing worst-case performance, defined as the average of the top quantile of the losses. In machine learning, such a risk measure is useful to train more robust models. Although the stochastic subgradient method (SGM) is a natural choice for minimizing the CVaR objective, we show that our stochastic prox-linear (SPL+) algorithm can better exploit the structure of the objective, while still providing a convenient closed form update. Our SPL+ method also adapts to the scaling of the loss function, which allows for easier tuning. We then specialize a general convergence theorem for SPL+ to our setting, and show that it allows for a wider selection of step sizes compared to SGM. We support this theoretical finding experimentally.

Instrumental variable (IV) methods are used to estimate causal effects in settings with unobserved confounding, where we cannot directly experiment on the treatment variable. Instruments are variables which only affect the outcome indirectly via the treatment variable(s). Most IV applications focus on low-dimensional treatments and crucially require at least as many instruments as treatments. This assumption is restrictive: in the natural sciences we often seek to infer causal effects of high-dimensional treatments (e.g., the effect of gene expressions or microbiota on health and disease), but can only run few experiments with a limited number of instruments (e.g., drugs or antibiotics). In such under-specified problems, the full treatment effect is not identifiable in a single experiment even in the linear case. We show that one can still reliably recover the projection of the treatment effect onto the instrumented subspace and develop techniques to consistently combine such partial estimates from different sets of instruments. We then leverage our combined estimators in an algorithm that iteratively proposes the most informative instruments at each round of experimentation to maximize the overall information about the full causal effect.

By incorporating regret minimization, double oracle methods have demonstrated rapid convergence to Nash Equilibrium (NE) in normal-form games and extensive-form games, through algorithms such as online double oracle (ODO) and extensive-form double oracle (XDO), respectively. In this study, we further examine the theoretical convergence rate and sample complexity of such regret minimization-based double oracle methods, utilizing a unified framework called Regret-Minimizing Double Oracle. Based on this framework, we extend ODO to extensive-form games and determine its sample complexity. Moreover, we demonstrate that the sample complexity of XDO can be exponential in the number of information sets $|S|$, owing to the exponentially decaying stopping threshold of restricted games. To solve this problem, we propose the Periodic Double Oracle (PDO) method, which has the lowest sample complexity among regret minimization-based double oracle methods, being only polynomial in $|S|$. Empirical evaluations on multiple poker and board games show that PDO achieves significantly faster convergence than previous double oracle algorithms and reaches a competitive level with state-of-the-art regret minimization methods.

In deep learning, often the training process finds an interpolator (a solution with 0 training loss), but the test loss is still low. This phenomenon, known as *benign overfitting*, is a major mystery that received a lot of recent attention. One common mechanism for benign overfitting is *implicit regularization*, where the training process leads to additional properties for the interpolator, often characterized by minimizing certain norms. However, even for a simple sparse linear regression problem $y = \beta^{\ast\top} x +\xi$ with sparse $\beta^{\ast}$, neither minimum $\ell_1$ or $\ell_2$ norm interpolator gives the optimal test loss. In this work, we give a different parametrization of the model which leads to a new implicit regularization effect that combines the benefit of $\ell_1$ and $\ell_2$ interpolators. We show that training our new model via gradient descent leads to an interpolator with near-optimal test loss. Our result is based on careful analysis of the training dynamics and provides another example of implicit regularization effect that goes beyond norm minimization.

Currently, most machine learning models are trained by centralized teams and are rarely updated. In contrast, open-source software development involves the iterative development of a shared artifact through distributed collaboration using a version control system. In the interest of enabling collaborative and continual improvement of machine learning models (Raffel, 2023), we introduce Git-Theta, a version control system for machine learning models. Git-Theta is an extension to Git, the most widely used version control software, that allows fine-grained tracking of changes to model parameters alongside code and other artifacts. Unlike existing version control systems that treat a model checkpoint as a blob of data, Git-Theta leverages the structure of checkpoints to support communication-efficient updates, automatic model merges, and meaningful reporting about the difference between two versions of a model. In addition, Git-Theta includes a plug-in system that enables users to easily add support for new functionality. In this paper, we introduce Git-Theta's design and features and include an example use-case of Git-Theta where a pre-trained model is continually adapted and modified. We publicly release Git-Theta in hopes of kickstarting a new era of collaborative model development. https://github.com/r-three/git-theta/

AI assistance continues to help advance applications in education, from language learning to intelligent tutoring systems, yet current methods for providing students feedback are still quite limited. Most automatic feedback systems either provide binary correctness feedback, which may not help a student understand how to improve, or require hand-coding feedback templates, which may not generalize to new domains. This can be particularly challenging for physical control tasks, where the rich diversity in student behavior and specialized domains make it challenging to leverage general-purpose assistive tools for providing feedback. We design and build CORGI, a model trained to generate language corrections for physical control tasks, such as learning to ride a bike. CORGI takes in as input a pair of student and expert trajectories, and then generates natural language corrections to help the student improve. We collect and train CORGI over data from three diverse physical control tasks (drawing, steering, and joint movement). Through both automatic and human evaluations, we show that CORGI can (i) generate valid feedback for novel student trajectories, (ii) outperform baselines on domains with novel control dynamics, and (iii) improve student learning in an interactive drawing task.

Given a matrix $A\in \mathbb{R}^{n\times d}$ and a vector $b\in \mathbb{R}^n$, we consider the regression problem with $\ell_\infty$ guarantees: finding a vector $x'\in \mathbb{R}^d$ such that $||x'-x^* ||_\infty \leq \frac{\epsilon}{\sqrt{d}}\cdot ||Ax^*-b||_2\cdot ||A^\dagger||$ with $x^*$ being the optimal solution to the regression $||Ax-b||_2$. One popular approach for solving $\ell_2$ regression problem is via sketching: picking a structured random matrix $S\in \mathbb{R}^{m\times n}$ with $m\ll n$ and $SA$ can be quickly computed, solve the ``sketched'' regression problem $x'=\mathrm{argmin} ||SAx-Sb||_2$. In this paper, we show that in order to obtain such $\ell_\infty$ guarantee for $\ell_2$ regression, one has to use sketching matrices that are *dense*. To the best of our knowledge, this is the first user case in which dense sketching matrices are necessary. On the algorithmic side, we prove that, there exists a distribution of dense sketching matrices with $m=\epsilon^{-2}d\log^3(n/\delta)$ such that solving the sketched regression problem gives the $\ell_\infty$ guarantee, with probability at least $1-\delta$. Moreover, the matrix $SA$ can be computed in time $O(nd\log n)$. Our row count is nearly-optimal up to logarithmic factors, and significantly improves the result in [Price, Song and Woodruff, ICALP'17], in which $m=\Omega(\epsilon^{-2}d^{1+\gamma})$ for $\gamma\in (0, 1)$ is required. Moreover, we develop a novel analytical framework for $\ell_\infty$ guarantee regression that utilizes the *Oblivious Coordinate-wise Embedding* (OCE) property introduced in [Song and Yu, ICML'21]. Our analysis is much simpler and more general than that of [Price, Song and Woodruff, ICALP'17]. Leveraging this framework, we extend the $\ell_\infty$ guarantee regression result to dense sketching matrices for computing fast tensor product of vectors.

Recently, contrastive learning has found impressive success in advancing the state of the art in solving various machine learning tasks. However, the existing generalization analysis is very limited or even not meaningful. In particular, the existing generalization error bounds depend linearly on the number $k$ of negative examples while it was widely shown in practice that choosing a large $k$ is necessary to guarantee good generalization of contrastive learning in downstream tasks. In this paper, we establish novel generalization bounds for contrastive learning which do not depend on $k$, up to logarithmic terms. Our analysis uses structural results on empirical covering numbers and Rademacher complexities to exploit the Lipschitz continuity of loss functions. For self-bounding Lipschitz loss functions, we further improve our results by developing optimistic bounds which imply fast rates in a low noise condition. We apply our results to learning with both linear representation and nonlinear representation by deep neural networks, for both of which we derive Rademacher complexity bounds to get improved generalization bounds.