<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head>
<body style="overflow-wrap: break-word; -webkit-nbsp-mode: space; line-break: after-white-space;">
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
Dear colleague,</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; min-height: 15px;">
<br>
</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
apologize for multiple posting</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; min-height: 15px;">
<br>
</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
<b>ScaDL 2023: Scalable Deep Learning over Parallel And Distributed</b></p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
<b>Infrastructure - An IPDPS 2023 Workshop</b></p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; min-height: 15px;">
<br>
</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; color: rgb(220, 161, 13);">
<span style="text-decoration: underline;"><a href="https://2023.scadl.org/">https://2023.scadl.org</a></span><span style="color: rgb(0, 0, 0);"> </span></p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; min-height: 15px;">
<br>
</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
<b>Scope of the Workshop:</b></p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
Recently, Deep Learning (DL) has received tremendous attention in the research</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
community because of the impressive results obtained for a large number of</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
machine learning problems. The success of state-of-the-art deep learning</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
systems relies on training deep neural networks over a massive amount of</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
training data, which typically requires a large-scale distributed computing</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
infrastructure to run. In order to run these jobs in a scalable and efficient</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
manner, on cloud infrastructure or dedicated HPC systems, several interesting</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
research topics have emerged which are specific to DL. The sheer size and</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
complexity of deep learning models when trained over a large amount of data</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
makes them harder to converge in a reasonable amount of time. It demands</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
advancement along multiple research directions such as, model/data</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
parallelism, model/data compression, distributed optimization algorithms for</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
DL convergence, synchronization strategies, efficient communication and</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
specific hardware acceleration.</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; min-height: 15px;">
<br>
</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
<b>SCADL seeks to advance the following research directions:</b></p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Asynchronous and Communication-Efficient SGD: Stochastic gradient descent is</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
at the core of large-scale machine learning. Parallelizing SGD gradient</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
computation across multiple nodes increases the data processed per iteration,</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
but exposes the SGD to communication and synchronization delays and</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
unpredictable node failures in the system. Thus, there is a critical need to</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
design robust and scalable distributed SGD methods to achieve fast error-</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
convergence in spite of such system variabilities.</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
High performance computing aspects: Deep learning is highly compute intensive.</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
Algorithms for kernel computations on commonly used accelerators (e.g. GPUs),</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
efficient techniques for communicating gradients and loading data from storage</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
are critical for training performance.</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; min-height: 15px;">
<br>
</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Model and Gradient Compression Techniques: Techniques such as reducing</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
weights and the size of weight tensors help in reducing the compute</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
complexity. Using lower-bit representations such as quantization and</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
sparsification allow for more optimal use of memory and communication</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
bandwidth.</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; min-height: 15px;">
<br>
</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Distributed Trustworthy AI: New techniques are needed to meet the goal of</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
global trustworthiness (e.g., fairness and adversarial robustness) efficiently</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
in a distributed DL setting.</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; min-height: 15px;">
<br>
</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Emerging AI hardware Accelerators: with the proliferation of new hardware</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
accelerators for AI such in memory computing (Analog AI) and neuromorphic</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
computing, novel methods and algorithms need to be introduced to adapt to the</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
underlying properties of the new hardware (example: the non-idealities of the</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
phase-change memory (PCM) and the cycle-to-cycle statistical variations).</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; min-height: 15px;">
<br>
</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- The intersection of Distributed DL and Neural Architecture Search (NAS): NAS</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
is increasingly being used to automate the synthesis of neural networks.</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
However, given the huge computational demands of NAS, distributed DL is</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
critical to make NAS computationally tractable (e.g., differentiable</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
distributed NAS).</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; min-height: 15px;">
<br>
</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
This intersection of distributed/parallel computing and deep learning is</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
becoming critical and demands specific attention to address the above topics</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
which some of the broader forums may not be able to provide. The aim of this</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
workshop is to foster collaboration among researchers from distributed/</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
parallel computing and deep learning communities to share the relevant topics</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
as well as results of the current approaches lying at the intersection of</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
these areas.</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; min-height: 15px;">
<br>
</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
<b>Areas of Interest</b></p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
In this workshop, we solicit research papers focused on distributed deep</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
learning aiming to achieve efficiency and scalability for deep learning jobs</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
over distributed and parallel systems. Papers focusing both on algorithms as</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
well as systems are welcome. We invite authors to submit papers on topics</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
including but not limited to:</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; min-height: 15px;">
<br>
</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Deep learning on cloud platforms, HPC systems, and edge devices</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Model-parallel and data-parallel techniques</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Asynchronous SGD for Training DNNs</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Communication-Efficient Training of DNNs</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Scalable and distributed graph neural networks, Sampling techniques for</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
graph neural networks</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Federated deep learning, both horizontal and vertical, and its challenges</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Model/data/gradient compression</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Learning in Resource constrained environments</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Coding Techniques for Straggler Mitigation</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Elasticity for deep learning jobs/spot market enablement</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Hyper-parameter tuning for deep learning jobs</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Hardware Acceleration for Deep Learning including digital and analog</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
accelerators</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Scalability of deep learning jobs on large clusters</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Deep learning on heterogeneous infrastructure</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Efficient and Scalable Inference</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Data storage/access in shared networks for deep learning</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Communication-efficient distributed fair and adversarially robust learning</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Distributed learning techniques applied to speed up neural architecture</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
search</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; min-height: 15px;">
<br>
</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; min-height: 15px;">
<br>
</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
<b>Workshop Format:</b></p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
Due to the continuing impact of COVID-19, ScaDL 2023 will also adopt relevant</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
IPDPS 2023 policies on virtual participation and presentation. Consequently,</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
the organizers are currently planning a hybrid (in-person and virtual) event.</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; min-height: 15px;">
<br>
</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
<b>Submission Link:</b></p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
Submissions will be managed through linklings. Submission link available at:</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; color: rgb(220, 161, 13);">
<span style="text-decoration: underline;"><a href="https://2023.scadl.org/call-for-papers">https://2023.scadl.org/call-for-papers</a></span><span style="color: rgb(0, 0, 0);"> </span></p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; min-height: 15px;">
<br>
</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
<b>Key Dates</b></p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
<b>Paper Submission: February 14th, 2023 (EXTENDED)</b></p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
<b>Acceptance Notification: February 26th, 2023</b></p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
<b>Camera ready papers due: March 3th, 2023</b></p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
<b>Workshop Date: May 19th, 2023 </b></p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; min-height: 15px;">
<br>
</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
<b>Author Instructions</b></p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
ScaDL 2023 accepts submissions in two categories:</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Regular papers: 8-10 pages</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
- Short papers/Work in progress: 4 pages</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
The aforementioned lengths include all technical content, references and</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
appendices.</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
We encourage submissions that are original research work, work in progress,</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
case studies, vision papers, and industrial experience papers.</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
Papers should be formatted using IEEE conference style, including figures,</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
tables, and references. The IEEE conference style templates for MS Word and</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
LaTeX provided by IEEE eXpress Conference Publishing are available for</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
download. See the latest versions at</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; color: rgb(220, 161, 13);">
<span style="text-decoration: underline;"><a href="https://www.ieee.org/conferences/publishing/templates.html">https://www.ieee.org/conferences/publishing/templates.html</a></span><span style="color: rgb(0, 0, 0);"> </span></p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; min-height: 15px;">
<br>
</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
<b>General Chairs</b></p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
Kaoutar El Maghraoui, IBM Research AI, USA</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
Daniele Lezzi, Barcelona Supercomputing Center, Spain</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; min-height: 15px;">
<br>
</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
<b>Program Committee Chairs</b></p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
Misbah Mubarak, NVIDIA, USA</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
Alex Gittens, Rensselaer Polytechnic Institute (RPI), USA</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; min-height: 15px;">
<br>
</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
<b>Publicity Chairs</b></p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
Federica Filippini, Politecnico di Milano, Italy</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
Hadjer Benmeziane, Université Polytechnique des Hauts-de-France</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; min-height: 15px;">
<br>
</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
<b>Web Chair </b></p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
Praveen Venkateswaran, IBM Research AI, USA</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue"; min-height: 15px;">
<br>
</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
<b>Steering Committee</b></p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
Parijat Dube, IBM Research AI, USA</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
Vinod Muthusamy, IBM Research AI, USA</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
Ashish Verma, IBM Research AI, USA</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
Jayaram K. R., IBM Research AI, USA</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
Yogish Sabharwal, IBM Research AI, India</p>
<p style="margin: 0px; font-stretch: normal; line-height: normal; font-family: "Helvetica Neue";">
Danilo Ardagna, Politecnico di Milano, Italy</p>
<br>
</body>
</html>