arXiv subjects

Hari Sundaram

Publications and source records attributed to Hari Sundaram.

At least 37 records · Page 2Linked to original sources

arXiv2025

AMUN: Adversarial Machine UNlearning

Machine unlearning, where users can request the deletion of a forget dataset, is becoming increasingly important because of numerous privacy regulations. Initial works on ``exact'' unlearning (e.g., retraining) incur large computational overheads. However, while computationally inexpensive, ``approximate'' methods have fallen short of reaching the effectiveness of exact unlearning: models produced fail to obtain comparable accuracy and prediction confidence on both the forget and test (i.e., unseen) dataset. Exploiting this observation, we propose a new unlearning method, Adversarial Machine UNlearning (AMUN), that outperforms prior state-of-the-art (SOTA) methods for image classification. AMUN lowers the confidence of the model on the forget samples by fine-tuning the model on their corresponding adversarial examples. Adversarial examples naturally belong to the distribution imposed by the model on the input space; fine-tuning the model on the adversarial examples closest to the corresponding forget samples (a) localizes the changes to the decision boundary of the model around each forget sample and (b) avoids drastic changes to the global behavior of the model, thereby preserving the model's accuracy on test samples. Using AMUN for unlearning a random $10\%$ of CIFAR-10 samples, we observe that even SOTA membership inference attacks cannot do better than random guessing.

cs.LG↗

arXiv2025

Do You Trust the Process?: Modeling Institutional Trust for Community Adoption of Reinforcement Learning Policies

Many governmental bodies are adopting AI policies for decision-making. In particular, Reinforcement Learning has been used to design policies that citizens would be expected to follow if implemented. Much RL work assumes that citizens follow these policies, and evaluate them with this in mind. However, we know from prior work that without institutional trust, citizens will not follow policies put in place by governments. In this work, we develop a trust-aware RL algorithm for resource allocation in communities. We consider the case of humanitarian engineering, where the organization is aiming to distribute some technology or resource to community members. We use a Deep Deterministic Policy Gradient approach to learn a resource allocation that fits the needs of the organization. Then, we simulate resource allocation according to the learned policy, and model the changes in institutional trust of community members. We investigate how this incorporation of institutional trust affects outcomes, and ask how effectively an organization can learn policies if trust values are private. We find that incorporating trust into RL algorithms can lead to more successful policies, specifically when the organization's goals are less certain. We find more conservative trust estimates lead to increased fairness and average community trust, though organization success suffers. Finally, we explore a strategy to prevent unfair outcomes to communities. We implement a quota system by an external entity which decreases the organization's utility when it does not serve enough community members. We find this intervention can improve fairness and trust among communities in some cases, while decreasing the success of the organization. This work underscores the importance of institutional trust in algorithm design and implementation, and identifies a tension between organization success and community well-being.

cs.LG↗

arXiv2025

Breaking Bad Tokens: Detoxification of LLMs Using Sparse Autoencoders

Large language models (LLMs) are now ubiquitous in user-facing applications, yet they still generate undesirable toxic outputs, including profanity, vulgarity, and derogatory remarks. Although numerous detoxification methods exist, most apply broad, surface-level fixes and can therefore easily be circumvented by jailbreak attacks. In this paper we leverage sparse autoencoders (SAEs) to identify toxicity-related directions in the residual stream of models and perform targeted activation steering using the corresponding decoder vectors. We introduce three tiers of steering aggressiveness and evaluate them on GPT-2 Small and Gemma-2-2B, revealing trade-offs between toxicity reduction and language fluency. At stronger steering strengths, these causal interventions surpass competitive baselines in reducing toxicity by up to 20%, though fluency can degrade noticeably on GPT-2 Small depending on the aggressiveness. Crucially, standard NLP benchmark scores upon steering remain stable, indicating that the model's knowledge and general abilities are preserved. We further show that feature-splitting in wider SAEs hampers safety interventions, underscoring the importance of disentangled feature learning. Our findings highlight both the promise and the current limitations of SAE-based causal interventions for LLM detoxification, further suggesting practical guidelines for safer language-model deployment.

cs.CL↗

arXiv2025

DIVEBATCH: Accelerating Model Training Through Gradient-Diversity Aware Batch Size Adaptation

The goal of this paper is to accelerate the training of machine learning models, a critical challenge since the training of large-scale deep neural models can be computationally expensive. Stochastic gradient descent (SGD) and its variants are widely used to train deep neural networks. In contrast to traditional approaches that focus on tuning the learning rate, we propose a novel adaptive batch size SGD algorithm, DiveBatch, that dynamically adjusts the batch size. Adapting the batch size is challenging: using large batch sizes is more efficient due to parallel computation, but small-batch training often converges in fewer epochs and generalizes better. To address this challenge, we introduce a data-driven adaptation based on gradient diversity, enabling DiveBatch to maintain the generalization performance of small-batch training while improving convergence speed and computational efficiency. Gradient diversity has a strong theoretical justification: it emerges from the convergence analysis of SGD. Evaluations of DiveBatch on synthetic and CiFar-10, CiFar-100, and Tiny-ImageNet demonstrate that DiveBatch converges significantly faster than standard SGD and AdaBatch (1.06 -- 5.0x), with a slight trade-off in performance.

cs.LG↗

arXiv2025

The Chilling: Identifying Strategic Antisocial Behavior Online and Examining the Impact on Journalists

On social platforms like Twitter, strategic targeted attacks are becoming increasingly common, especially against vulnerable groups such as female journalists. Two key challenges in identifying strategic online behavior are the complex structure of online conversations and the hidden nature of potential strategies that drive user behavior. To address these, we develop a new tree structured Transformer model that categorizes replies based on their hierarchical conversation structures. Extensive experiments demonstrate that our proposed classification model can effectively detect different user groups, namely attackers, supporters, and bystanders, and their latent strategies. To demonstrate the utility of our approach, we apply this classifier to real time Twitter data and conduct a series of quantitative analyses on the interactions between journalists with different groups of users. Our classification approach allows us to not only explore strategic behaviors of attackers but also those of supporters and bystanders who engage in online interactions. When examining the impact of online attacks, we find a strong correlation between the presence of attackers' interactions and chilling effects, where journalists tend to slow their subsequent posting behavior. This paper provides a deeper understanding of how different user groups engage in online discussions and highlights the detrimental effects of attacker presence on journalists, other users, and conversational outcomes. Our findings underscore the need for social platforms to develop tools that address coordinated toxicity. By detecting patterns of coordinated attacks early, platforms could limit the visibility of toxic content to prevent escalation. Additionally, providing journalists and users with tools for real time reporting could empower them to manage hostile interactions more effectively.

cs.SI↗

arXiv2025

LLM-RecG: A Semantic Bias-Aware Framework for Zero-Shot Sequential Recommendation

Zero-shot cross-domain sequential recommendation (ZCDSR) enables predictions in unseen domains without additional training or fine-tuning, addressing the limitations of traditional models in sparse data environments. Recent advancements in large language models (LLMs) have significantly enhanced ZCDSR by facilitating cross-domain knowledge transfer through rich, pretrained representations. Despite this progress, domain semantic bias -- arising from differences in vocabulary and content focus between domains -- remains a persistent challenge, leading to misaligned item embeddings and reduced generalization across domains. To address this, we propose a novel semantic bias-aware framework that enhances LLM-based ZCDSR by improving cross-domain alignment at both the item and sequential levels. At the item level, we introduce a generalization loss that aligns the embeddings of items across domains (inter-domain compactness), while preserving the unique characteristics of each item within its own domain (intra-domain diversity). This ensures that item embeddings can be transferred effectively between domains without collapsing into overly generic or uniform representations. At the sequential level, we develop a method to transfer user behavioral patterns by clustering source domain user sequences and applying attention-based aggregation during target domain inference. We dynamically adapt user embeddings to unseen domains, enabling effective zero-shot recommendations without requiring target-domain interactions...

cs.IR↗

arXiv2025

A Pre-trained Sequential Recommendation Framework: Popularity Dynamics for Zero-shot Transfer

Sequential recommenders are crucial to the success of online applications, \eg e-commerce, video streaming, and social media. While model architectures continue to improve, for every new application domain, we still have to train a new model from scratch for high quality recommendations. On the other hand, pre-trained language and vision models have shown great success in zero-shot or few-shot adaptation to new application domains. Inspired by the success of pre-trained models in peer AI fields, we propose a novel pre-trained sequential recommendation framework: PrepRec. We learn universal item representations by modeling item popularity dynamics. Through extensive experiments on five real-world datasets, we show that PrepRec, without any auxiliary information, can not only zero-shot transfer to a new domain, but achieve competitive performance compared to state-of-the-art sequential recommender models with only a fraction of the model size. In addition, with a simple post-hoc interpolation, PrepRec can improve the performance of existing sequential recommenders on average by 13.8\% in Recall@10 and 29.5% in NDCG@10. We provide an anonymized implementation of PrepRec at https://anonymous.4open.science/r/PrepRec--2F60/

cs.IR↗

arXiv2025

Organize, Then Vote: Exploring Cognitive Load in Quadratic Survey Interfaces

Quadratic Surveys (QSs) elicit more accurate preferences than traditional methods like Likert-scale surveys. However, the cognitive load associated with QSs has hindered their adoption in digital surveys for collective decision-making. We introduce a two-phase "organize-then-vote" QS to reduce cognitive load. As interface design significantly impacts survey results and accuracy, our design scaffolds survey takers' decision-making while managing the cognitive load imposed by QS. In a 2x2 between-subject in-lab study on public resource allotment, we compared our interface with a traditional text interface across a QS with 6 (short) and 24 (long) options. Two-phase interface participants spent more time per option and exhibited shorter voting edit distances. We qualitatively observed shifts in cognitive effort from mechanical operations to constructing more comprehensive preferences. We conclude that this interface promoted deeper engagement, potentially reducing satisficing behaviors caused by cognitive overload in longer QSs. This research clarifies how human-centered design improves preference elicitation tools for collective decision-making.

cs.HC↗

arXiv2025

Oh the Prices You'll See: Designing a Fair Exchange System to Mitigate Personalized Pricing

Many online marketplaces personalize prices based on consumer attributes. Since these prices are private, consumers may be unaware that they have spent more on a good than the lowest possible price, and cannot easily take action to pay less. In this paper, we introduce a fairness-centered exchange system that takes advantage of personalized pricing, while still allowing consumers to individually benefit. Our system produces a matching of consumers to promote trading; the lower-paying consumer buys the good for the higher-paying consumer for some fee. We explore various modeling choices and fairness targets to determine which schema will leave consumers best off, while also earning revenue for the system itself. We show that when consumers individually negotiate the transaction price, and our fairness objective is to minimize mean net cost, we are able to achieve the most fair outcomes. Conversely, when transaction prices are centrally set, consumers are often unwilling to transact. When price dispersion (or range) is high, the system can reduce the mean net cost to each individual by $66\%$, or the mean net cost to a group by $69\%$. We find that a high dispersion of original prices is necessary for our system to be viable. Higher dispersion can actually lead to decreased net price paid by consumers, and act as a check against extreme personalization, increasing seller accountability. Our results provide theoretical evidence that such a system could improve fairness for consumers while sustaining itself financially.

cs.GT↗

arXiv2025

Algorithmic Collective Action with Two Collectives

Given that data-dependent algorithmic systems have become impactful in more domains of life, the need for individuals to promote their own interests and hold algorithms accountable has grown. To have meaningful influence, individuals must band together to engage in collective action. Groups that engage in such algorithmic collective action are likely to vary in size, membership characteristics, and crucially, objectives. In this work, we introduce a first of a kind framework for studying collective action with two or more collectives that strategically behave to manipulate data-driven systems. With more than one collective acting on a system, unexpected interactions may occur. We use this framework to conduct experiments with language model-based classifiers and recommender systems where two collectives each attempt to achieve their own individual objectives. We examine how differing objectives, strategies, sizes, and homogeneity can impact a collective's efficacy. We find that the unintentional interactions between collectives can be quite significant; a collective acting in isolation may be able to achieve their objective (e.g., improve classification outcomes for themselves or promote a particular item), but when a second collective acts simultaneously, the efficacy of the first group drops by as much as $75\%$. We find that, in the recommender system context, neither fully heterogeneous nor fully homogeneous collectives stand out as most efficacious and that heterogeneity's impact is secondary compared to collective size. Our results signal the need for more transparency in both the underlying algorithmic models and the different behaviors individuals or collectives may take on these systems. This approach also allows collectives to hold algorithmic system developers accountable and provides a framework for people to actively use their own data to promote their own interests.

cs.CY↗

arXiv2025

Enhancing Computational Notebooks with Code+Data Space Versioning

There is a gap between how people explore data and how Jupyter-like computational notebooks are designed. People explore data nonlinearly, using execution undos, branching, and/or complete reverts, whereas notebooks are designed for sequential exploration. Recent works like ForkIt are still insufficient to support these multiple modes of nonlinear exploration in a unified way. In this work, we address the challenge by introducing two-dimensional code+data space versioning for computational notebooks and verifying its effectiveness using our prototype system, Kishuboard, which integrates with Jupyter. By adjusting code and data knobs, users of Kishuboard can intuitively manage the state of computational notebooks in a flexible way, thereby achieving both execution rollbacks and checkouts across complex multi-branch exploration history. Moreover, this two-dimensional versioning mechanism can easily be presented along with a friendly one-dimensional history. Human subject studies indicate that Kishuboard significantly enhances user productivity in various data science tasks.

cs.HC↗

arXiv2025

Large-scale Evaluation of Notebook Checkpointing with AI Agents

Saving, or checkpointing, intermediate results during interactive data exploration can potentially boost user productivity. However, existing studies on this topic are limited, as they primarily rely on small-scale experiments with human participants - a fundamental constraint of human subject studies. To address this limitation, we employ AI agents to simulate a large number of complex data exploration scenarios, including revisiting past states and branching into new exploration paths. This strategy enables us to accurately assess the impact of checkpointing while closely mimicking the behavior of real-world data practitioners. Our evaluation results, involving more than 1,000 exploration paths and 2,848 executed code blocks, show that a checkpointing framework for computational notebooks can indeed enhance productivity by minimizing unnecessary code re-executions and redundant variables or code.

cs.HC↗

arXiv2024

Venire: A Machine Learning-Guided Panel Review System for Community Content Moderation

Research into community content moderation often assumes that moderation teams govern with a single, unified voice. However, recent work has found that moderators disagree with one another at modest, but concerning rates. The problem is not the root disagreements themselves. Subjectivity in moderation is unavoidable, and there are clear benefits to including diverse perspectives within a moderation team. Instead, the crux of the issue is that, due to resource constraints, moderation decisions end up being made by individual decision-makers. The result is decision-making that is inconsistent, which is frustrating for community members. To address this, we develop Venire, an ML-backed system for panel review on Reddit. Venire uses a machine learning model trained on log data to identify the cases where moderators are most likely to disagree. Venire fast-tracks these cases for multi-person review. Ideally, Venire allows moderators to surface and resolve disagreements that would have otherwise gone unnoticed. We conduct three studies through which we design and evaluate Venire: a set of formative interviews with moderators, technical evaluations on two datasets, and a think-aloud study in which moderators used Venire to make decisions on real moderation cases. Quantitatively, we demonstrate that Venire is able to improve decision consistency and surface latent disagreements. Qualitatively, we find that Venire helps moderators resolve difficult moderation cases more confidently. Venire represents a novel paradigm for human-AI content moderation, and shifts the conversation from replacing human decision-making to supporting it.

cs.HC↗

arXiv2024

Spectrum Extraction and Clipping for Implicitly Linear Layers

We show the effectiveness of automatic differentiation in efficiently and correctly computing and controlling the spectrum of implicitly linear operators, a rich family of layer types including all standard convolutional and dense layers. We provide the first clipping method which is correct for general convolution layers, and illuminate the representational limitation that caused correctness issues in prior work. We study the effect of the batch normalization layers when concatenated with convolutional layers and show how our clipping method can be applied to their composition. By comparing the accuracy and performance of our algorithms to the state-of-the-art methods, using various experiments, we show they are more precise and efficient and lead to better generalization and adversarial robustness. We provide the code for using our methods at https://github.com/Ali-E/FastClip.

cs.LG↗

arXiv2024

LOTOS: Layer-wise Orthogonalization for Training Robust Ensembles

Transferability of adversarial examples is a well-known property that endangers all classification models, even those that are only accessible through black-box queries. Prior work has shown that an ensemble of models is more resilient to transferability: the probability that an adversarial example is effective against most models of the ensemble is low. Thus, most ongoing research focuses on improving ensemble diversity. Another line of prior work has shown that Lipschitz continuity of the models can make models more robust since it limits how a model's output changes with small input perturbations. In this paper, we study the effect of Lipschitz continuity on transferability rates. We show that although a lower Lipschitz constant increases the robustness of a single model, it is not as beneficial in training robust ensembles as it increases the transferability rate of adversarial examples across models in the ensemble. Therefore, we introduce LOTOS, a new training paradigm for ensembles, which counteracts this adverse effect. It does so by promoting orthogonality among the top-$k$ sub-spaces of the transformations of the corresponding affine layers of any pair of models in the ensemble. We theoretically show that $k$ does not need to be large for convolutional layers, which makes the computational overhead negligible. Through various experiments, we show LOTOS increases the robust accuracy of ensembles of ResNet-18 models by $6$ percentage points (p.p) against black-box attacks on CIFAR-10. It is also capable of combining with the robustness of prior state-of-the-art methods for training robust ensembles to enhance their robust accuracy by $10.7$ p.p.

cs.LG↗

arXiv2024

CodeScope: An Execution-based Multilingual Multitask Multidimensional Benchmark for Evaluating LLMs on Code Understanding and Generation

Large Language Models (LLMs) have demonstrated remarkable performance on assisting humans in programming and facilitating programming automation. However, existing benchmarks for evaluating the code understanding and generation capacities of LLMs suffer from severe limitations. First, most benchmarks are insufficient as they focus on a narrow range of popular programming languages and specific tasks, whereas real-world software development scenarios show a critical need to implement systems with multilingual and multitask programming environments to satisfy diverse requirements. Second, most benchmarks fail to consider the actual executability and the consistency of execution results of the generated code. To bridge these gaps between existing benchmarks and expectations from practical applications, we introduce CodeScope, an execution-based, multilingual, multitask, multidimensional evaluation benchmark for comprehensively measuring LLM capabilities on coding tasks. CodeScope covers 43 programming languages and eight coding tasks. It evaluates the coding performance of LLMs from three dimensions (perspectives): length, difficulty, and efficiency. To facilitate execution-based evaluations of code generation, we develop MultiCodeEngine, an automated code execution engine that supports 14 programming languages. Finally, we systematically evaluate and analyze eight mainstream LLMs and demonstrate the superior breadth and challenges of CodeScope for evaluating LLMs on code understanding and generation tasks compared to other benchmarks. The CodeScope benchmark and code are publicly available at https://github.com/WeixiangYAN/CodeScope.

cs.CL↗

arXiv2024

CEV-LM: Controlled Edit Vector Language Model for Shaping Natural Language Generations

As large-scale language models become the standard for text generation, there is a greater need to tailor the generations to be more or less concise, targeted, and informative, depending on the audience/application. Existing control approaches primarily adjust the semantic (e.g., emotion, topics), structural (e.g., syntax tree, parts-of-speech), and lexical (e.g., keyword/phrase inclusion) properties of text, but are insufficient to accomplish complex objectives such as pacing which control the complexity and readability of the text. In this paper, we introduce CEV-LM - a lightweight, semi-autoregressive language model that utilizes constrained edit vectors to control three complementary metrics (speed, volume, and circuitousness) that quantify the shape of text (e.g., pacing of content). We study an extensive set of state-of-the-art CTG models and find that CEV-LM provides significantly more targeted and precise control of these three metrics while preserving semantic content, using less training data, and containing fewer parameters.

cs.CL↗

arXiv2024

Advancing Precise Outline-Conditioned Text Generation with Task Duality and Explicit Outline Control

Existing works on outline-conditioned text generation typically aim to generate text using provided outlines as rough sketches, such as keywords and phrases. However, these approaches make it challenging to control the quality of text generation and assess consistency between outlines and generated texts due to lack of clarity and rationality of the rough outlines. In this paper, we introduce a novel text generation task called Precise Outline-conditioned Generation, which requires generating stories based on specific, sentence-level outlines. To facilitate research on this task, we construct two new datasets, WPOG and CDM. We provide strong baselines based on fine-tuning models such as BART and GPT-2, and evaluating zero-shot performance of models such as ChatGPT and Vicuna. Furthermore, we identify an issue of imbalanced utilization of the outline information in the precise outline-conditioned generation, which is ubiquitously observed across fine-tuned models and zero-shot inference models. To address this issue, we propose an explicit outline utilization control approach and a novel framework that leverages the task duality between summarization and generation. Experimental results show that the proposed approaches effectively alleviate the issue of imbalanced outline utilization and enhance the quality of precise outline-conditioned text generation for both fine-tuning and zero-shot settings.

cs.CL↗