arXiv subjects

Ashish Mishra

Publications and source records attributed to Ashish Mishra.

At least 19 recordsLinked to original sources

arXiv2026

Liquid Tree Automata

Component-based synthesis (CBS) generates loop-free programs from library components to satisfy logical queries. While expressive specifications and precise queries simplify the solution space, they make finding feasible execution paths significantly more difficult for traditional CBS procedures. As constraints become more exact, the search must navigate an increasingly sparse space of valid paths. We address this challenge by reasoning about \emph{logical similarities} between exploration paths. We consider library methods equipped with refinement-type specifications, which enrich base types with logical qualifiers to precisely constrain the value space. To efficiently explore this space, we introduce Liquid Tree Automata (LTA), a novel tree automata variant whose construction is driven by refinement typing rules. LTAs leverage subtyping constraints to identify and eagerly merge semantically similar states during search. This merging avoids redundant exploration of equivalent paths, significantly improving synthesis efficiency. We implement this approach in a tool called Hegel. Our evaluation demonstrates that Hegel synthesizes solutions to complex queries that are beyond the reach of existing state-of-the-art tools.

cs.PL↗

arXiv2026

On factorization of matrix of Kazhdan-Lusztig polynomials

Let $\mathcal{H} = \mathcal{H}(W,S)$ be the Hecke algebra of the Coxeter system $(W,S)$ over $\mathbb{Z}[q^{\pm1}]$, where $W$ is the Weyl group of a symmetrizable Kac-Moody algebra. In this paper, we show that the matrix of Kazhdan-Lusztig polynomials of $\mathcal{H}$ factorizes into a product of $|S|$ many matrices, each of which has entries as polynomials in $q$ with nonnegative coefficients. To achieve this goal, we use hybrid basis $TC^J$ for $J\subseteq S$ of $\mathcal{H}$, defined by Grojnowski-Haiman. The intermediate matrices in the aforementioned factorization turn out to be the transition matrices from $TC^J$-basis to $TC^I$-basis for $I\subset J$. Equivalently, these coefficients can be computed using a natural restriction map from $\mathcal{H}$ to the parabolic Hecke algebra $\mathcal{H}_J$. Moreover, following the ideas from Grojnowski-Haiman, we also give a geometric proof of the positivity of these coefficients.

math.RT↗

arXiv2025

Selective, Controlled and Domain-Agnostic Unlearning in Pretrained CLIP: A Training- and Data-Free Approach

Pretrained models like CLIP have demonstrated impressive zero-shot classification capabilities across diverse visual domains, spanning natural images, artistic renderings, and abstract representations. However, real-world applications often demand the removal (or "unlearning") of specific object classes without requiring additional data or retraining, or affecting the model's performance on unrelated tasks. In this paper, we propose a novel training- and data-free unlearning framework that enables three distinct forgetting paradigms: (1) global unlearning of selected objects across all domains, (2) domain-specific knowledge removal (e.g., eliminating sketch representations while preserving photo recognition), and (3) complete unlearning in selective domains. By leveraging a multimodal nullspace through synergistic integration of text prompts and synthesized visual prototypes derived from CLIP's joint embedding space, our method efficiently removes undesired class information while preserving the remaining knowledge. This approach overcomes the limitations of existing retraining-based methods and offers a flexible and computationally efficient solution for controlled model forgetting.

cs.CV↗

arXiv2025

Erasing CLIP Memories: Non-Destructive, Data-Free Zero-Shot class Unlearning in CLIP Models

We introduce a novel, closed-form approach for selective unlearning in multimodal models, specifically targeting pretrained models such as CLIP. Our method leverages nullspace projection to erase the target class information embedded in the final projection layer, without requiring any retraining or the use of images from the forget set. By computing an orthonormal basis for the subspace spanned by target text embeddings and projecting these directions, we dramatically reduce the alignment between image features and undesired classes. Unlike traditional unlearning techniques that rely on iterative fine-tuning and extensive data curation, our approach is both computationally efficient and surgically precise. This leads to a pronounced drop in zero-shot performance for the target classes while preserving the overall multimodal knowledge of the model. Our experiments demonstrate that even a partial projection can balance between complete unlearning and retaining useful information, addressing key challenges in model decontamination and privacy preservation.

cs.CV↗

arXiv2025

Forecast2Anomaly (F2A): Adapting Multivariate Time Series Foundation Models for Anomaly Prediction

Forecasting anomalies (anomaly prediction) in multivariate time series from different real-world, dynamic, and complex systems is vital for preempting critical failures, leading to a substantial minimization in operational costs and human labor. Yet, existing methods are limited to specific systems while failing to generalize to evolving anomaly patterns over time. In contrast, pretrained Time Series Foundation Models (TSFMs) have recently demonstrated strong generalization and zero-shot forecasting capabilities. However, their potential remains untapped for anomaly prediction, a task fundamentally different from forecasting normal behavior. Thus, we present Forecast2Anomaly (F2A), a novel framework that empowers TSFMs with anomaly prediction abilities through two key innovations. First, we propose a joint forecast-anomaly loss that fine-tunes TSFMs to accurately forecast future signals even at anomalous time points. Second, we introduce a Retrieval-Augmented Generation (RAG) module that retrieves historically relevant horizons and conditions predictions on them. This component dynamically adapts to distributional shifts at inference time, enabling F2A to track evolving anomalies without requiring model updates. By combining targeted fine-tuning with dynamic retrieval, F2A bridges the gap between robust TSFM zero-shot forecasting and zero-shot anomaly prediction. Extensive experiments across 16 diverse datasets and multiple TSFM backbones show that F2A consistently outperforms state-of-the-art methods, offering a scalable, zero-shot anomaly prediction solution for real-world applications.

cs.LG↗

arXiv2025

Close is Good Enough: Component-Based Synthesis Modulo Logical Similarity

Component-based synthesis (CBS) aims to generate loop-free programs from a set of libraries whose methods are annotated with specifications and whose output must satisfy a set of logical constraints, expressed as a query. The effectiveness of a CBS algorithm critically depends on the severity of the constraints imposed by the query. The more exact these constraints are, the sparser the space of feasible solutions. This maxim also applies when we enrich the expressiveness of the specifications affixed to library methods. In both cases, the search must now contend with constraints that may only hold over a small number of the possible execution paths that can be enumerated by a CBS procedure. In this paper, we address this challenge by equipping CBS search with the ability to reason about logical similarities among the paths it explores. Our setting considers library methods equipped with refinement-type specifications that enrich ordinary base types with a set of rich logical qualifiers to constrain the set of values accepted by that type. We perform a search over a tree automata variant called Qualified Tree Automata that intelligently records information about enumerated terms, leveraging subtyping constraints over the refinement types associated with these terms to enable reasoning about similarity among candidate solutions as search proceeds, thereby avoiding exploration of semantically similar paths. We present an implementation of this idea in a tool called \name, and provide a comprehensive evaluation that demonstrates \name's ability to synthesize solutions to complex CBS queries that go well-beyond the capabilities of the existing state-of-the-art.

cs.PL↗

arXiv2025

We've Got You Covered: Type-Guided Repair of Incomplete Input Generators

Property-based testing (PBT) is a popular technique for automatically testing semantic properties of a program, specified as a pair of pre- and post-conditions. The efficacy of this approach depends on being able to quickly generate inputs that meet the precondition, in order to maximize the set of program behaviors that are probed. For semantically rich preconditions, purely random generation is unlikely to produce many valid inputs; when this occurs, users are forced to manually write their own specialized input generators. One common problem with handwritten generators is that they may be incomplete, i.e., they are unable to generate some values meeting the target precondition. This paper presents a novel program repair technique that patches an incomplete generator so that its range includes every valid input. Our approach uses a novel enumerative synthesis algorithm that leverages the recently developed notion of coverage types to characterize the set of missing test values as well as the coverage provided by candidate repairs. We have implemented a repair tool for OCaml generators, called Cobb, and used it to repair a suite of benchmarks drawn from the PBT literature.

cs.PL↗

arXiv2024

BadScan: An Architectural Backdoor Attack on Visual State Space Models

The newly introduced Visual State Space Model (VMamba), which employs \textit{State Space Mechanisms} (SSM) to interpret images as sequences of patches, has shown exceptional performance compared to Vision Transformers (ViT) across various computer vision tasks. However, recent studies have highlighted that deep models are susceptible to adversarial attacks. One common approach is to embed a trigger in the training data to retrain the model, causing it to misclassify data samples into a target class, a phenomenon known as a backdoor attack. In this paper, we first evaluate the robustness of the VMamba model against existing backdoor attacks. Based on this evaluation, we introduce a novel architectural backdoor attack, termed BadScan, designed to deceive the VMamba model. This attack utilizes bit plane slicing to create visually imperceptible backdoored images. During testing, if a trigger is detected by performing XOR operations between the $k^{th}$ bit planes of the modified triggered patches, the traditional 2D selective scan (SS2D) mechanism in the visual state space (VSS) block of VMamba is replaced with our newly designed BadScan block, which incorporates four newly developed scanning patterns. We demonstrate that the BadScan backdoor attack represents a significant threat to visual state space models and remains effective even after complete retraining from scratch. Experimental results on two widely used image classification datasets, CIFAR-10, and ImageNet-1K, reveal that while visual state space models generally exhibit robustness against current backdoor attacks, the BadScan attack is particularly effective, achieving a higher Triggered Accuracy Ratio (TAR) in misleading the VMamba model and its variants.

cs.CV↗

arXiv2023

Morpheus: Automated Safety Verification of Data-dependent Parser Combinator Programs

Parser combinators are a well-known mechanism used for the compositional construction of parsers, and have shown to be particularly useful in writing parsers for rich grammars with data-dependencies and global state. Verifying applications written using them, however, has proven to be challenging in large part because of the inherently effectful nature of the parsers being composed and the difficulty in reasoning about the arbitrarily rich data-dependent semantic actions that can be associated with parsing actions. In this paper, we address these challenges by defining a parser combinator framework called Morpheus equipped with abstractions for defining composable effects tailored for parsing and semantic actions and a rich specification language used to define safety properties over the constituent parsers comprising a program. Even though its abstractions yield many of the same expressivity benefits as other parser combinator systems, Morpheus is carefully engineered to yield a substantially more tractable automated verification pathway. We demonstrate its utility in verifying a number of realistic, challenging parsing applications, including several cases that involve non-trivial data-dependent relations.

cs.PL↗

arXiv2023

Polymorphic Coverage Types

Test input generators are an important part of property-based testing (PBT) frameworks, and a key expectation is that they be capable of producing all acceptable elements that satisfy both the function's input type and the generator-provided constraints. However, it is not readily apparent how to validate whether a particular generator's output satisfies this coverage requirement. In practice, developers typically rely on manual inspection and post-mortem analysis of test runs to determine whether a generator provides sufficient coverage; these approaches are error-prone and difficult to scale as generators grow more complex. To address this problem, we present a new refinement-type-based verification procedure for validating the coverage provided by input test generators, based on a novel interpretation of types that embeds "must-style" underapproximate reasoning principles as a fundamental part of the type system. In our formulation, the types associated with expressions capture the set of values guaranteed to be produced by the expression, rather than the usual interpretation in which types represent the set of values an expression may produce. We formalize the notion of coverage types in a rich core language supporting higher-order functions and inductive datatypes. To better support real-world test generators, we extend this type system with type and qualifier polymorphism, enabling static verification of coverage guarantees for test input generators constructed using the monadic combinators found in most PBT frameworks. Finally, we have implemented a coverage type checker for OCaml programs based on this core calculus and present a detailed evaluation of its utility using a corpus of benchmarks drawn from both the PBT literature and open-source projects.

cs.PL↗

arXiv2022

Specification-Guided Component-Based Synthesis from Effectful Libraries

Component-based synthesis seeks to build programs using the APIs provided by a set of libraries. Oftentimes, these APIs have effects, which make it challenging to reason about the correctness of potential synthesis candidates. This is because changes to global state made by effectful library procedures affect how they may be composed together, yielding an intractably large search space that can confound typical enumerative synthesis techniques. If the nature of these effects are exposed as part of their specification, however, deductive synthesis approaches can be used to help guide the search for components. In this paper, we present a new specification-guided synthesis procedure that uses Hoare-style pre- and post-conditions to express fine-grained effects of potential library component candidates to drive a bi-directional synthesis search strategy. The procedure alternates between a forward search process that seeks to build larger terms given an existing context but which is otherwise unaware of the actual goal, alongside a backward search mechanism that seeks terms consistent with the desired goal but which is otherwise unaware of the context from which these terms must be synthesized. To further improve efficiency and scalability, we integrate a conflict-driven learning procedure into the synthesis algorithm that provides a semantic characterization of previously encountered unsuccessful search paths that is used to prune the space of possible candidates as synthesis proceeds. We have implemented our ideas in a tool called Cobalt and demonstrate its effectiveness on a number of challenging synthesis problems defined over OCaml libraries equipped with effectful specifications.

cs.PL↗

arXiv2022

On Quasi Steinberg characters of Complex Reflection Groups

Let $G$ be a finite group and $p$ be a prime number dividing the order of $G$. An irreducible character $χ$ of $G$ is called a quasi $p$-Steinberg character if $χ(g)$ is nonzero for every $p$-regular element $g$ in $G$. In this paper, we classify quasi $p$-Steinberg characters of the complex reflection groups $G(r,q,n)$. In particular, we obtain this classification for Weyl groups of type $B_n$ and type $D_n$.

math.RT↗

arXiv2021

Stubbifier: Debloating Dynamic Server-Side JavaScript Applications

JavaScript is an increasingly popular language for server-side development, thanks in part to the Node.js runtime environment and its vast ecosystem of modules. With the Node.js package manager npm, users are able to easily include external modules as dependencies in their projects. However, npm installs modules with all of their functionality, even if only a fraction is needed, which causes an undue increase in code size. Eliminating this unused functionality from distributions is desirable, but the sound analysis required to find unused code is difficult due to JavaScript's extreme dynamicity. We present a fully automatic technique that identifies unused code by constructing static or dynamic call graphs from the application's tests, and replacing code deemed unreachable with either file- or function-level stubs. If a stub is called, it will fetch and execute the original code on-demand, thus relaxing the requirement that the call graph be sound. The technique also provides an optional guarded execution mode to guard application against injection vulnerabilities in untested code that resulted from stub expansion. This technique is implemented in an open source tool called Stubbifier, which supports the ECMAScript 2019 standard. In an empirical evaluation on 15 Node.js applications and 75 clients of these applications, Stubbifier reduced application size by 56% on average while incurring only minor performance overhead. The evaluation also shows that Stubbifier's guarded execution mode is capable of preventing several known injection vulnerabilities that are manifested in stubbed-out code. Finally, Stubbifier can work alongside bundlers, popular JavaScript tools for bundling an application with its dependencies. For the considered subject applications, we measured an average size reduction of 37% in bundled distributions.

cs.SE↗

arXiv2021

Outbursts and stellar properties of the classical Be star HD 6226

The bright and understudied classical Be star HD 6226 has exhibited multiple outbursts in the last several years during which the star grew a viscous decretion disk. We analyze 659 optical spectra of the system collected from 2017-2020, along with a UV spectrum from the Hubble Space Telescope and high cadence photometry from both TESS and the KELT survey. We find that the star has a spectral type of B2.5IIIe, with a rotation rate of 74% of critical. The star is nearly pole-on with an inclination of $13.4$ degree. We confirm the spectroscopic pulsational properties previously reported, and report on three photometric oscillations from KELT photometry. The outbursting behavior is studied with equivalent width measurements of H$α$ and H$β$, and the variations in both of these can be quantitatively explained with two frequencies through a Fourier analysis. One of the frequencies for the emission outbursts is equal to the difference between two photometric oscillations, linking these pulsation modes to the mass ejection mechanism for some outbursts. During the TESS observation time period of 2019 October 7 to 2019 November 2, the star was building a disk. With a large dataset of H$α$ and H$β$ spectroscopy, we are able to determine the timescales of dissipation in both of these lines, similar to past work on Be stars that has been done with optical photometry. HD 6226 is an ideal target with which to study the Be disk-evolution given its apparent periodic nature, allowing for targeted observations with other facilities in the future.

astro-ph.SR↗

arXiv2021

Generative Replay-based Continual Zero-Shot Learning

Zero-shot learning is a new paradigm to classify objects from classes that are not available at training time. Zero-shot learning (ZSL) methods have attracted considerable attention in recent years because of their ability to classify unseen/novel class examples. Most of the existing approaches on ZSL works when all the samples from seen classes are available to train the model, which does not suit real life. In this paper, we tackle this hindrance by developing a generative replay-based continual ZSL (GRCZSL). The proposed method endows traditional ZSL to learn from streaming data and acquire new knowledge without forgetting the previous tasks' gained experience. We handle catastrophic forgetting in GRCZSL by replaying the synthetic samples of seen classes, which have appeared in the earlier tasks. These synthetic samples are synthesized using the trained conditional variational autoencoder (VAE) over the immediate past task. Moreover, we only require the current and immediate previous VAE at any time for training and testing. The proposed GRZSL method is developed for a single-head setting of continual learning, simulating a real-world problem setting. In this setting, task identity is given during training but unavailable during testing. GRCZSL performance is evaluated on five benchmark datasets for the generalized setup of ZSL with fixed and dynamic (incremental class) settings of continual learning. The existing class setting presented recently in the literature is not suitable for a class-incremental setting. Therefore, this paper proposes a new setting to address this issue. Experimental results show that the proposed method significantly outperforms the baseline and the state-of-the-art method and makes it more suitable for real-world applications.

cs.CV↗

arXiv2021

Online Lifelong Generalized Zero-Shot Learning

Methods proposed in the literature for zero-shot learning (ZSL) are typically suitable for offline learning and cannot continually learn from sequential streaming data. The sequential data comes in the form of tasks during training. Recently, a few attempts have been made to handle this issue and develop continual ZSL (CZSL) methods. However, these CZSL methods require clear task-boundary information between the tasks during training, which is not practically possible. This paper proposes a task-free (i.e., task-agnostic) CZSL method, which does not require any task information during continual learning. The proposed task-free CZSL method employs a variational autoencoder (VAE) for performing ZSL. To develop the CZSL method, we combine the concept of experience replay with knowledge distillation and regularization. Here, knowledge distillation is performed using the training sample's dark knowledge, which essentially helps overcome the catastrophic forgetting issue. Further, it is enabled for task-free learning using short-term memory. Finally, a classifier is trained on the synthetic features generated at the latent space of the VAE. Moreover, the experiments are conducted in a challenging and practical ZSL setup, i.e., generalized ZSL (GZSL). These experiments are conducted for two kinds of single-head continual learning settings: (i) mild setting-: task-boundary is known only during training but not during testing; (ii) strict setting-: task-boundary is not known at training, as well as testing. Experimental results on five benchmark datasets exhibit the validity of the approach for CZSL.

cs.CV↗

arXiv2021

Generalized Continual Zero-Shot Learning

Recently, zero-shot learning (ZSL) emerged as an exciting topic and attracted a lot of attention. ZSL aims to classify unseen classes by transferring the knowledge from seen classes to unseen classes based on the class description. Despite showing promising performance, ZSL approaches assume that the training samples from all seen classes are available during the training, which is practically not feasible. To address this issue, we propose a more generalized and practical setup for ZSL, i.e., continual ZSL (CZSL), where classes arrive sequentially in the form of a task and it actively learns from the changing environment by leveraging the past experience. Further, to enhance the reliability, we develop CZSL for a single head continual learning setting where task identity is revealed during the training process but not during the testing. To avoid catastrophic forgetting and intransigence, we use knowledge distillation and storing and replay the few samples from previous tasks using a small episodic memory. We develop baselines and evaluate generalized CZSL on five ZSL benchmark datasets for two different settings of continual learning: with and without class incremental. Moreover, CZSL is developed for two types of variational autoencoders, which generates two types of features for classification: (i) generated features at output space and (ii) generated discriminative features at the latent space. The experimental results clearly indicate the single head CZSL is more generalizable and suitable for practical applications.

cs.CV↗

arXiv2020

EMPOL: an EMCCD based optical imaging polarimeter

An Andor 1K $\times$ 1K EMCCD detector has been used to develop an optical imaging polarimeter for use at the Cassegrain focus of 1.2 m telescope of PRL. The optics is derived from an older single-element detector instrument and consists of a rotating half-wave plate as modulator and a Foster prism as an analyser. The field of view of the instrument is 3 $\times$ 3 sq arcmin. We describe the instrument and the observational methodology in this document. Extensive observations have been carried out with this instrument covering a large variety of sources e.g. near-Earth asteroids, comets, Lynds dark nebulae, open clusters and AGN such as blazars. In the current communication, we discuss some results from the initial calibration runs while the other results will be presented elsewhere.

astro-ph.IM↗