arXiv subjects

Lisa Zhang

Publications and source records attributed to Lisa Zhang.

13 recordsLinked to original sources

arXiv2026

AI-Generated Slides: Are They Good? Can Students Tell?

As generative AI (GenAI) tools become easily accessible, there is promise in using such tools to support instructors. To that end, this paper examines using GenAI to help generate slides from instructor authored course notes, emphasizing instructor and student perceptions. We examine an end-to-end education tool (NotebookLM), two general-purpose LLMs (Claude, M365 Copilot), and two coding assistants (Cursor, Claude Code). We first analyze whether GenAI generated slides are ``good'' via narrative assessment by educators. We choose the best slides to use (with some modification) in a real course setting, and compare the student perception of human vs. AI generated slides. We find that coding assistant tools produce slides that were most accurate, complete, and pedagogically sound. Additionally, students rate GenAI slides to be of similar quality as instructor-created slides, and cannot reliably identify which slides are AI-generated. Additionally, we find a negative correlation between a high quality rating and a high ``AI-generated'' rating, suggesting students associate poor quality with the source of the slides being AI. These findings highlight promising opportunities for integrating GenAI into instructional design workflows and call for further research on how educators can best harness such tools responsibly and effectively.

cs.AI↗

arXiv2024

Understanding Help-Seeking Behavior of Students Using LLMs vs. Web Search for Writing SQL Queries

Growth in the use of large language models (LLMs) in programming education is altering how students write SQL queries. Traditionally, students relied heavily on web search for coding assistance, but this has shifted with the adoption of LLMs like ChatGPT. However, the comparative process and outcomes of using web search versus LLMs for coding help remain underexplored. To address this, we conducted a randomized interview study in a database classroom to compare web search and LLMs, including a publicly available LLM (ChatGPT) and an instructor-tuned LLM, for writing SQL queries. Our findings indicate that using an instructor-tuned LLM required significantly more interactions than both ChatGPT and web search, but resulted in a similar number of edits to the final SQL query. No significant differences were found in the quality of the final SQL queries between conditions, although the LLM conditions directionally showed higher query quality. Furthermore, students using instructor-tuned LLM reported a lower mental demand. These results have implications for learning and productivity in programming education.

cs.HC↗

arXiv2024

Decomposed Prompting to Answer Questions on a Course Discussion Board

We propose and evaluate a question-answering system that uses decomposed prompting to classify and answer student questions on a course discussion board. Our system uses a large language model (LLM) to classify questions into one of four types: conceptual, homework, logistics, and not answerable. This enables us to employ a different strategy for answering questions that fall under different types. Using a variant of GPT-3, we achieve $81\%$ classification accuracy. We discuss our system's performance on answering conceptual questions from a machine learning course and various failure modes.

cs.CL↗

arXiv2019

Reviving and Improving Recurrent Back-Propagation

In this paper, we revisit the recurrent back-propagation (RBP) algorithm, discuss the conditions under which it applies as well as how to satisfy them in deep neural networks. We show that RBP can be unstable and propose two variants based on conjugate gradient on the normal equations (CG-RBP) and Neumann series (Neumann-RBP). We further investigate the relationship between Neumann-RBP and back propagation through time (BPTT) and its truncated version (TBPTT). Our Neumann-RBP has the same time complexity as TBPTT but only requires constant memory, whereas TBPTT's memory cost scales linearly with the number of truncation steps. We examine all RBP variants along with BPTT and TBPTT in three different application domains: associative memory with continuous Hopfield networks, document classification in citation networks using graph neural networks and hyperparameter optimization for fully connected networks. All experiments demonstrate that RBPs, especially the Neumann-RBP variant, are efficient and effective for optimizing convergent recurrent neural networks. Code is released at: \url{https://github.com/lrjconan/RBP}.

cs.LG↗

arXiv2019

Inference in Probabilistic Graphical Models by Graph Neural Networks

A fundamental computation for statistical inference and accurate decision-making is to compute the marginal probabilities or most probable states of task-relevant variables. Probabilistic graphical models can efficiently represent the structure of such complex data, but performing these inferences is generally difficult. Message-passing algorithms, such as belief propagation, are a natural way to disseminate evidence amongst correlated variables while exploiting the graph structure, but these algorithms can struggle when the conditional dependency graphs contain loops. Here we use Graph Neural Networks (GNNs) to learn a message-passing algorithm that solves these inference tasks. We first show that the architecture of GNNs is well-matched to inference tasks. We then demonstrate the efficacy of this inference approach by training GNNs on a collection of graphical models and showing that they substantially outperform belief propagation on loopy graphs. Our message-passing algorithms generalize out of the training set to larger graphs and graphs with different structure.

cs.LG↗

arXiv2018

Neural Guided Constraint Logic Programming for Program Synthesis

Synthesizing programs using example input/outputs is a classic problem in artificial intelligence. We present a method for solving Programming By Example (PBE) problems by using a neural model to guide the search of a constraint logic programming system called miniKanren. Crucially, the neural model uses miniKanren's internal representation as input; miniKanren represents a PBE problem as recursive constraints imposed by the provided examples. We explore Recurrent Neural Network and Graph Neural Network models. We contribute a modified miniKanren, drivable by an external agent, available at https://github.com/xuexue/neuralkanren. We show that our neural-guided approach using constraints can synthesize programs faster in many cases, and importantly, can generalize to larger problems.

cs.LG↗

arXiv2018

Learning deep structured active contours end-to-end

The world is covered with millions of buildings, and precisely knowing each instance's position and extents is vital to a multitude of applications. Recently, automated building footprint segmentation models have shown superior detection accuracy thanks to the usage of Convolutional Neural Networks (CNN). However, even the latest evolutions struggle to precisely delineating borders, which often leads to geometric distortions and inadvertent fusion of adjacent building instances. We propose to overcome this issue by exploiting the distinct geometric properties of buildings. To this end, we present Deep Structured Active Contours (DSAC), a novel framework that integrates priors and constraints into the segmentation process, such as continuous boundaries, smooth edges, and sharp corners. To do so, DSAC employs Active Contour Models (ACM), a family of constraint- and prior-based polygonal models. We learn ACM parameterizations per instance using a CNN, and show how to incorporate all components in a structured output model, making DSAC trainable end-to-end. We evaluate DSAC on three challenging building instance segmentation datasets, where it compares favorably against state-of-the-art. Code will be made available.

cs.CV↗

arXiv2017

Egalitarian Graph Orientations

Given an undirected graph, one can assign directions to each of the edges of the graph, thus orienting the graph. To be as egalitarian as possible, one may wish to find an orientation such that no vertex is unfairly hit with too many arcs directed into it. We discuss how this objective arises in problems resulting from telecommunications. We give optimal, polynomial-time algorithms for: finding an orientation that minimizes the lexicographic order of the indegrees and finding a strongly-connected orientation that minimizes the maximum indegree. We show that minimizing the lexicographic order of the indegrees is NP-hard when the resulting orientation is required to be acyclic.

cs.DM↗

arXiv2014

Improving Robustness of Next-Hop Routing

A weakness of next-hop routing is that following a link or router failure there may be no routes between some source-destination pairs, or packets may get stuck in a routing loop as the protocol operates to establish new routes. In this article, we address these weaknesses by describing mechanisms to choose alternate next hops. Our first contribution is to model the scenario as the following {\sc tree augmentation} problem. Consider a mixed graph where some edges are directed and some undirected. The directed edges form a spanning tree pointing towards the common destination node. Each directed edge represents the unique next hop in the routing protocol. Our goal is to direct the undirected edges so that the resulting graph remains acyclic and the number of nodes with outdegree two or more is maximized. These nodes represent those with alternative next hops in their routing paths. We show that {\sc tree augmentation} is NP-hard in general and present a simple $\frac{1}{2}$-approximation algorithm. We also study 3 special cases. We give exact polynomial-time algorithms for when the input spanning tree consists of exactly 2 directed paths or when the input graph has bounded treewidth. For planar graphs, we present a polynomial-time approximation scheme when the input tree is a breadth-first search tree. To the best of our knowledge, {\sc tree augmentation} has not been previously studied.

cs.DS↗

arXiv2013

Utility Optimization in Heterogeneous Networks via CSMA-Based Algorithms

We study algorithms for carrier and rate allocation in cellular systems with distributed components such as a heterogeneous LTE system with macrocells and femtocells. Existing work on LTE systems often involves centralized techniques or requires significant signaling, and is therefore not always applicable in the presence of femtocells. More distributed CSMA-based algorithms (carrier-sense multiple access) were developed in the context of 802.11 systems and have been proven to be utility optimal. However, the proof typically assumes a single transmission rate on each carrier. Further, it relies on the CSMA collision detection mechanisms to know whether a transmission is feasible. In this paper we present a framework for LTE scheduling that is based on CSMA techniques. In particular we first prove that CSMA-based algorithms can be generalized to handle multiple transmission rates in a multi-carrier setting while maintaining utility optimality. We then show how such an algorithm can be implemented in a heterogeneous LTE system where the existing Channel Quality Indication (CQI) mechanism is used to decide transmission feasibility.

cs.NI↗

arXiv2013

Energy-Delay Tradeoffs in a Load-Balanced Router

The Load-Balanced Router architecture has received a lot of attention because it does not require centralized scheduling at the internal switch fabrics. In this paper we reexamine the architecture, motivated by its potential to turn off multiple components and thereby conserve energy in the presence of low traffic. We perform a detailed analysis of the queue and delay performance of a Load-Balanced Router under a simple random routing algorithm. We calculate probabilistic bounds for queue size and delay, and show that the probabilities drop exponentially with increasing queue size or delay. We also demonstrate a tradeoff in energy consumption against the queue and delay performance.

cs.NI↗

arXiv2010

Limits of Approximation Algorithms: PCPs and Unique Games (DIMACS Tutorial Lecture Notes)

These are the lecture notes for the DIMACS Tutorial "Limits of Approximation Algorithms: PCPs and Unique Games" held at the DIMACS Center, CoRE Building, Rutgers University on 20-21 July, 2009. This tutorial was jointly sponsored by the DIMACS Special Focus on Hardness of Approximation, the DIMACS Special Focus on Algorithmic Foundations of the Internet, and the Center for Computational Intractability with support from the National Security Agency and the National Science Foundation. The speakers at the tutorial were Matthew Andrews, Sanjeev Arora, Moses Charikar, Prahladh Harsha, Subhash Khot, Dana Moshkovitz and Lisa Zhang. The sribes were Ashkan Aazami, Dev Desai, Igor Gorodezky, Geetha Jagannathan, Alexander S. Kulikov, Darakhshan J. Mir, Alantha Newman, Aleksandar Nikolov, David Pritchard and Gwen Spencer.

cs.CC↗

arXiv2002

Source Routing and Scheduling in Packet Networks

We study {\em routing} and {\em scheduling} in packet-switched networks. We assume an adversary that controls the injection time, source, and destination for each packet injected. A set of paths for these packets is {\em admissible} if no link in the network is overloaded. We present the first on-line routing algorithm that finds a set of admissible paths whenever this is feasible. Our algorithm calculates a path for each packet as soon as it is injected at its source using a simple shortest path computation. The length of a link reflects its current congestion. We also show how our algorithm can be implemented under today's Internet routing paradigms. When the paths are known (either given by the adversary or computed as above) our goal is to schedule the packets along the given paths so that the packets experience small end-to-end delays. The best previous delay bounds for deterministic and distributed scheduling protocols were exponential in the path length. In this paper we present the first deterministic and distributed scheduling protocol that guarantees a polynomial end-to-end delay for every packet. Finally, we discuss the effects of combining routing with scheduling. We first show that some unstable scheduling protocols remain unstable no matter how the paths are chosen. However, the freedom to choose paths can make a difference. For example, we show that a ring with parallel links is stable for all greedy scheduling protocols if paths are chosen intelligently, whereas this is not the case if the adversary specifies the paths.

cs.NI↗