arXiv subjects

Usama Khalid

Publications and source records attributed to Usama Khalid.

7 recordsLinked to original sources

arXiv2026

You Can't Prefer Emotions You Don't Sample: Intensity Undershoot in DPO-Tuned LLMs

Ask a language model to respond "very excitedly," and its output is typically only mildly more energetic. We quantify this effect. We condition an instruction-tuned LLM on a continuous Valence-Arousal (VA) target, where valence measures how pleasant a state is and arousal how activated it is, measure the achieved affect with a frozen regressor, and sweep the requested target from -1 to +1. The response moves far less than asked: the gain, the slope of achieved against requested affect, is only 0.26 for valence and 0.13 for arousal on Llama-3.1-8B, where a faithful controller would score 1. The model systematically undershoots requested emotional intensity, which puts a number on the qualitative observation of Fazzi et al. (2025). Our experiments trace this to the preference-learning pipeline. Training targets from natural corpora such as EmoBank are neutral-heavy, and the sampled candidates themselves rarely reach extreme affect, so Direct Preference Optimization (DPO) is left with no extreme exemplar to prefer. If instead we cover the target space uniformly and sample a hotter, larger candidate pool, valence gain rises from 0.26 to 0.40 +/- 0.02 (3 seeds) and extrapolation error drops, at only a modest in-distribution cost (EmoBank-test VA distance 0.092 to 0.107). The same recipe reproduces on Qwen3-8B (gain_v 0.44, with in-distribution accuracy preserved). Arousal is harder and less reliable: its gain barely moves on average and swings across seeds (0.14 +/- 0.07, against valence's tight +/- 0.02), because raising arousal needs candidates the base model is reluctant to generate. The evidence indicates that faithful intensity is bottlenecked by the extremity of the candidate pool rather than by the conditioning format.

cs.CL↗

arXiv2024

Innovative SiC ultraviolet instrumentation development with potential applications for the Habitable Worlds Observatory

In this paper, we detail recent and current work that is being carried out to fabricate and advance novel SiC UV instrumentation that is aimed at enabling more sensitive measurements across numerous disciplines, with a short discussion of the promise such detectors may hold for the Habitable Worlds Observatory. We discuss SiC instrument development progress that is being carried out under multiple NASA grants, including several PICASSO and SBIR grants, as well as an ECI grant. Testing of pixel design, properties and layout as well as maturation of the integration scheme developed through these efforts provide key technology and engineering advancement for potential HWO detectors. Achieving desired noise characteristics, responsivity, and validating operation of SiC detectors using standard read out techniques offers a compelling platform for operation of denser and higher dimensionality SiC photodiode arrays of interest for use in potential HWO Coronagraph, Spectrograph, and High Resolution Imaging Instruments. We incorporate these SiC detector properties into a simulation of potential NUV exoplanet observations by HWO using SiC detectors and also discuss potential application to HWO.

astro-ph.IM↗

arXiv2021

Few Shot Learning for Information Verification

Information verification is quite a challenging task, this is because many times verifying a claim can require picking pieces of information from multiple pieces of evidence which can have a hierarchy of complex semantic relations. Previously a lot of researchers have mainly focused on simply concatenating multiple evidence sentences to accept or reject claims. These approaches are limited as evidence can contain hierarchical information and dependencies. In this research, we aim to verify facts based on evidence selected from a list of articles taken from Wikipedia. Pretrained language models such as XLNET are used to generate meaningful representations and graph-based attention and convolutions are used in such a way that the system requires little additional training to learn to verify facts.

cs.CL↗

arXiv2021

Co-occurrences using Fasttext embeddings for word similarity tasks in Urdu

Urdu is a widely spoken language in South Asia. Though immoderate literature exists for the Urdu language still the data isn't enough to naturally process the language by NLP techniques. Very efficient language models exist for the English language, a high resource language, but Urdu and other under-resourced languages have been neglected for a long time. To create efficient language models for these languages we must have good word embedding models. For Urdu, we can only find word embeddings trained and developed using the skip-gram model. In this paper, we have built a corpus for Urdu by scraping and integrating data from various sources and compiled a vocabulary for the Urdu language. We also modify fasttext embeddings and N-Grams models to enable training them on our built corpus. We have used these trained embeddings for a word similarity task and compared the results with existing techniques.

cs.CL↗

arXiv2021

Bilingual Language Modeling, A transfer learning technique for Roman Urdu

Pretrained language models are now of widespread use in Natural Language Processing. Despite their success, applying them to Low Resource languages is still a huge challenge. Although Multilingual models hold great promise, applying them to specific low-resource languages e.g. Roman Urdu can be excessive. In this paper, we show how the code-switching property of languages may be used to perform cross-lingual transfer learning from a corresponding high resource language. We also show how this transfer learning technique termed Bilingual Language Modeling can be used to produce better performing models for Roman Urdu. To enable training and experimentation, we also present a collection of novel corpora for Roman Urdu extracted from various sources and social networking sites, e.g. Twitter. We train Monolingual, Multilingual, and Bilingual models of Roman Urdu - the proposed bilingual model achieves 23% accuracy compared to the 2% and 11% of the monolingual and multilingual models respectively in the Masked Language Modeling (MLM) task.

cs.CL↗

arXiv2021

RUBERT: A Bilingual Roman Urdu BERT Using Cross Lingual Transfer Learning

In recent studies, it has been shown that Multilingual language models underperform their monolingual counterparts. It is also a well-known fact that training and maintaining monolingual models for each language is a costly and time-consuming process. Roman Urdu is a resource-starved language used popularly on social media platforms and chat apps. In this research, we propose a novel dataset of scraped tweets containing 54M tokens and 3M sentences. Additionally, we also propose RUBERT a bilingual Roman Urdu model created by additional pretraining of English BERT. We compare its performance with a monolingual Roman Urdu BERT trained from scratch and a multilingual Roman Urdu BERT created by additional pretraining of Multilingual BERT. We show through our experiments that additional pretraining of the English BERT produces the most notable performance improvement.

cs.CL↗

arXiv2020

Exploiting ML algorithms for Efficient Detection and Prevention of JavaScript-XSS Attacks in Android Based Hybrid Applications

The development and analysis of mobile applications in term of security have become an active research area from many years as many apps are vulnerable to different attacks. Especially the concept of hybrid applications has emerged in the last three years where applications are developed in both native and web languages because the use of web languages raises certain security risks in hybrid mobile applications as it creates possible channels where malicious code can be injected inside the application. WebView is an important component in hybrid mobile applications which used to implements a sandbox mechanism to protect the local resources of smartphone devices from un-authorized access of JavaScript. However, the WebView application program interfaces (APIs) also have security issues. For example, an attacker can attack the hybrid application via JavaScript code by bypassing the sandbox security through accessing the public methods of the applications. Cross-site scripting (XSS) is one of the most popular malicious code injection technique for accessing the public methods of the application through JavaScript. This research proposes a framework for detection and prevention of XSS attacks in hybrid applications using state-of-the-art machine learning (ML) algorithms. The detection of the attacks have been perform by exploiting the registered Java object features. The dataset and the sample hybrid applications have been developed using the android studio. Then the widely used toolkit, RapidMiner, has been used for empirical analysis. The results reveal that the ensemble based Random Forest algorithm outperforms other algorithms and achieves both the accuracy and F-measures as high as of 99%.

cs.CR↗