Author name cluster

Amit Deshpande

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

10 papers

2 author rows

NeurIPS Conference 2025 Conference Paper

On Optimal Steering to Achieve Exact Fairness

Mohit Sharma
Amit Deshpande
Chiranjib Bhattacharyya
Rajiv Ratn Shah

To fix the `bias in, bias out' problem in fair machine learning, it is important to steer feature distributions of data or internal representations of Large Language Models (LLMs) to \emph{ideal} ones that guarantee group-fair outcomes. Previous work on fair generative models and representation steering could greatly benefit from provable fairness guarantees on the model output. We define a distribution as \emph{ideal} if the minimizer of any cost-sensitive risk on it is guaranteed to have exact group-fair outcomes (e. g. , demographic parity, equal opportunity)---in other words, it has no fairness-utility trade-off. We formulate an optimization program for optimal steering by finding the nearest \emph{ideal} distribution in KL-divergence, and provide efficient algorithms for it when the underlying distributions come from well-known parametric families (e. g. , normal, log-normal). Empirically, our optimal steering techniques on both synthetic and real-world datasets improve fairness without diminishing utility (and sometimes even improve utility). We demonstrate affine steering of LLM representations to reduce bias in multi-class classification, e. g. , occupation prediction from a short biography in Bios dataset (De-Arteaga et al. ). Furthermore, we steer internal representations of LLMs towards desired outputs so that it works equally well across different groups.

PDF Details

AAAI Conference 2024 Conference Paper

Rethinking Robustness of Model Attributions

Sandesh Kamath
Sankalp Mittal
Amit Deshpande
Vineeth N Balasubramanian

For machine learning models to be reliable and trustworthy, their decisions must be interpretable. As these models find increasing use in safety-critical applications, it is important that not just the model predictions but also their explanations (as feature attributions) be robust to small human-imperceptible input perturbations. Recent works have shown that many attribution methods are fragile and have proposed improvements in either these methods or the model training. We observe two main causes for fragile attributions: first, the existing metrics of robustness (e.g., top-k intersection) overpenalize even reasonable local shifts in attribution, thereby making random perturbations to appear as a strong attack, and second, the attribution can be concentrated in a small region even when there are multiple important parts in an image. To rectify this, we propose simple ways to strengthen existing metrics and attribution methods that incorporate locality of pixels in robustness metrics and diversity of pixel locations in attributions. Towards the role of model training in attributional robustness, we empirically observe that adversarially trained models have more robust attributions on smaller datasets, however, this advantage disappears in larger datasets. Code is made available at https://github.com/ksandeshk/LENS.

PDF Details DOI

NeurIPS Conference 2023 Conference Paper

Causal Effect Regularization: Automated Detection and Removal of Spurious Correlations

Abhinav Kumar
Amit Deshpande
Amit Sharma

In many classification datasets, the task labels are spuriously correlated with some input attributes. Classifiers trained on such datasets often rely on these attributes for prediction, especially when the spurious correlation is high, and thus fail togeneralize whenever there is a shift in the attributes’ correlation at deployment. If we assume that the spurious attributes are known a priori, several methods have been proposed to learn a classifier that is invariant to the specified attributes. However, in real-world data, information about spurious attributes is typically unavailable. Therefore, we propose a method that automatically identifies spurious attributes by estimating their causal effect on the label and then uses a regularization objective to mitigate the classifier’s reliance on them. Although causal effect of an attribute on the label is not always identified, we present two commonly occurring data-generating processes where the effect can be identified. Compared to recent work for identifying spurious attributes, we find that our method, AutoACER, ismore accurate in removing the attribute from the learned model, especially when spurious correlation is high. Specifically, across synthetic, semi-synthetic, and real-world datasets, AutoACER shows significant improvement in a metric used to quantify the dependence of a classifier on spurious attributes ($\Delta$Prob), while obtaining better or similar accuracy. Empirically we find that AutoACER mitigatesthe reliance on spurious attributes even under noisy estimation of causal effects or when the causal effect is not identified. To explain the empirical robustness of our method, we create a simple linear classification task with two sets of attributes: causal and spurious. Under this setting, we prove that AutoACER only requires the ranking of estimated causal effects to be correct across attributes to select thecorrect classifier.

PDF Details

IJCAI Conference 2023 Conference Paper

Sampling Ex-Post Group-Fair Rankings

Sruthi Gorantla
Amit Deshpande
Anand Louis

Randomized rankings have been of recent interest to achieve ex-ante fairer exposure and better robustness than deterministic rankings. We propose a set of natural axioms for randomized group-fair rankings and prove that there exists a unique distribution D that satisfies our axioms and is supported only over ex-post group-fair rankings, i. e. , rankings that satisfy given lower and upper bounds on group-wise representation in the top-k ranks. Our problem formulation works even when there is implicit bias, incomplete relevance information, or only ordinal ranking is available instead of relevance scores or utility values. We propose two algorithms to sample a random group-fair ranking from the distribution D mentioned above. Our first dynamic programming-based algorithm samples ex-post group-fair rankings uniformly at random in time O(k^2 ell), where "ell" is the number of groups. Our second random walk-based algorithm samples ex-post group-fair rankings from a distribution epsilon-close to D in total variation distance and has expected running time O*(k^2 ell^2), when there is a sufficient gap between the given upper and lower bounds on the group-wise representation. The former does exact sampling, but the latter runs significantly faster on real-world data sets for larger values of k. We give empirical evidence that our algorithms compare favorably against recent baselines for fairness and ranking utility on real-world data sets.

PDF Details DOI

NeurIPS Conference 2021 Conference Paper

Can we have it all? On the Trade-off between Spatial and Adversarial Robustness of Neural Networks

Sandesh Kamath
Amit Deshpande
Subrahmanyam Kambhampati Venkata
Vineeth N Balasubramanian

(Non-)robustness of neural networks to small, adversarial pixel-wise perturbations, and as more recently shown, to even random spatial transformations (e. g. , translations, rotations) entreats both theoretical and empirical understanding. Spatial robustness to random translations and rotations is commonly attained via equivariant models (e. g. , StdCNNs, GCNNs) and training augmentation, whereas adversarial robustness is typically achieved by adversarial training. In this paper, we prove a quantitative trade-off between spatial and adversarial robustness in a simple statistical setting. We complement this empirically by showing that: (a) as the spatial robustness of equivariant models improves by training augmentation with progressively larger transformations, their adversarial robustness worsens progressively, and (b) as the state-of-the-art robust models are adversarially trained with progressively larger pixel-wise perturbations, their spatial robustness drops progressively. Towards achieving Pareto-optimality in this trade-off, we propose a method based on curriculum learning that trains gradually on more difficult perturbations (both spatial and adversarial) to improve spatial and adversarial robustness simultaneously.

PDF Details

AAAI Conference 2021 Conference Paper

How Linguistically Fair Are Multilingual Pre-Trained Language Models?

Monojit Choudhury
Amit Deshpande

Massively multilingual pre-trained language models, such as mBERT and XLM-RoBERTa, have received significant attention in the recent NLP literature for their excellent capability towards crosslingual zero-shot transfer of NLP tasks. This is especially promising because a large number of languages have no or very little labeled data for supervised learning. Moreover, a substantially improved performance on low resource languages without any significant degradation of accuracy for high resource languages lead us to believe that these models will help attain a fairer distribution of language technologies despite the prevalent unfair and extremely skewed distribution of resources across the world’s languages. Nevertheless, these models, and the experimental approaches adopted by the researchers to arrive at those, have been criticised by some for lacking a nuanced and thorough comparison of benefits across languages and tasks. A related and important question that has received little attention is how to choose from a set of models, when no single model significantly outperforms the others on all tasks and languages. As we discuss in this paper, this is often the case, and the choices are usually made without a clear articulation of reasons or underlying fairness assumptions. In this work, we scrutinize the choices made in previous work, and propose a few different strategies for fair and efficient model selection based on the principles of fairness in economics and social choice theory. In particular, we emphasize Rawlsian fairness, which provides an appropriate framework for making fair (with respect to languages, or tasks, or both) choices while selecting multilingual pre-trained language models for a practical or scientific set-up.

PDF Details

ICML Conference 2021 Conference Paper

On the Problem of Underranking in Group-Fair Ranking

Sruthi Gorantla
Amit Deshpande
Anand Louis

Bias in ranking systems, especially among the top ranks, can worsen social and economic inequalities, polarize opinions, and reinforce stereotypes. On the other hand, a bias correction for minority groups can cause more harm if perceived as favoring group-fair outcomes over meritocracy. Most group-fair ranking algorithms post-process a given ranking and output a group-fair ranking. In this paper, we formulate the problem of underranking in group-fair rankings based on how close the group-fair rank of each item is to its original rank, and prove a lower bound on the trade-off achievable for simultaneous underranking and group fairness in ranking. We give a fair ranking algorithm that takes any given ranking and outputs another ranking with simultaneous underranking and group fairness guarantees comparable to the lower bound we prove. Our experimental results confirm the theoretical trade-off between underranking and group fairness, and also show that our algorithm achieves the best of both when compared to the state-of-the-art baselines.

Details

TCS Journal 2021 Journal Article

Sampling-based dimension reduction for subspace approximation with outliers

Amit Deshpande
Rameshwar Pratap

The subspace approximation problem with outliers, for given n points in d dimensions x 1, x 2, …, x n ∈ R d, an integer 1 ≤ k ≤ d, and an outlier parameter 0 ≤ α ≤ 1, is to find a k-dimensional linear subspace of R d that minimizes the sum of squared distances to its nearest ( 1 − α ) n points. More generally, the ℓ p subspace approximation problem with outliers minimizes the sum of p-th powers of distances instead of the sum of squared distances. Even the case of p = 2 or robust PCA is non-trivial, and previous work requires additional assumptions on the input or generative models for it. Any multiplicative approximation algorithm for the subspace approximation problem with outliers must solve the robust subspace recovery problem, a special case in which the ( 1 − α ) n inliers in the optimal solution are promised to lie exactly on a k-dimensional linear subspace. However, robust subspace recovery is Small Set Expansion (SSE)-hard, and known algorithmic results for robust subspace recovery require strong assumptions on the input, e. g. , any d outliers must be linearly independent. In this paper, we show how to extend dimension reduction techniques and bi-criteria approximations based on sampling and coresets to the problem of subspace approximation with outliers. To get around the SSE-hardness of robust subspace recovery, we assume that the squared distance error of the optimal k-dimensional subspace summed over the optimal ( 1 − α ) n inliers is at least δ times its squared-error summed over all n points, for some 0 < δ ≤ 1 − α. Under this assumption, we give an efficient algorithm to find a weak coreset or a subset of poly ( k / ϵ ) log ⁡ ( 1 / δ ) log ⁡ log ⁡ ( 1 / δ ) points whose span contains a k-dimensional subspace that gives a multiplicative ( 1 + ϵ ) -approximation to the optimal solution. Our technique is based on the squared-length sampling algorithm suggested for low-rank approximation problems in the seminal work of Frieze, Kannan, and Vempala [12]. The running time of our algorithm is linear in n and d. Interestingly, our results hold even when the fraction of outliers α is large, as long as the obvious condition 0 < δ ≤ 1 − α is satisfied. We show similar results for subspace approximation with ℓ p error or more general M-estimator loss functions, and also give an additive approximation for the affine subspace approximation problem.

Details DOI

AAAI Conference 2021 Conference Paper

The Importance of Modeling Data Missingness in Algorithmic Fairness: A Causal Perspective

Naman Goel
Alfonso Amayuelas
Amit Deshpande
Amit Sharma

Training datasets for machine learning often have some form of missingness. For example, to learn a model for deciding whom to give a loan, the available training data includes individuals who were given a loan in the past, but not those who were not. This missingness, if ignored, nullifies any fairness guarantee of the training procedure when the model is deployed. Using causal graphs, we characterize the missingness mechanisms in different real-world scenarios. We show conditions under which various distributions, used in popular fairness algorithms, can or can not be recovered from the training data. Our theoretical results imply that many of these algorithms can not guarantee fairness in practice. Modeling missingness also helps to identify correct design principles for fair algorithms. For example, in multi-stage settings where decisions are made in multiple screening rounds, we use our framework to derive the minimal distributions required to design a fair algorithm. Our proposed algorithm decentralizes the decision-making process and still achieves similar performance to the optimal algorithm that requires centralization and non-recoverable distributions.

PDF Details

NeurIPS Conference 2016 Conference Paper

Batched Gaussian Process Bandit Optimization via Determinantal Point Processes

Tarun Kathuria
Amit Deshpande
Pushmeet Kohli

Gaussian Process bandit optimization has emerged as a powerful tool for optimizing noisy black box functions. One example in machine learning is hyper-parameter optimization where each evaluation of the target function may require training a model which may involve days or even weeks of computation. Most methods for this so-called “Bayesian optimization” only allow sequential exploration of the parameter space. However, it is often desirable to propose batches or sets of parameter values to explore simultaneously, especially when there are large parallel processing facilities at our disposal. Batch methods require modeling the interaction between the different evaluations in the batch, which can be expensive in complex scenarios. In this paper, we propose a new approach for parallelizing Bayesian optimization by modeling the diversity of a batch via Determinantal point processes (DPPs) whose kernels are learned automatically. This allows us to generalize a previous result as well as prove better regret bounds based on DPP sampling. Our experiments on a variety of synthetic and real-world robotics and hyper-parameter optimization tasks indicate that our DPP-based methods, especially those based on DPP sampling, outperform state-of-the-art methods.

PDF Details