<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head>
<body>
<div dir="auto">Yeah, that sounds good to me.</div>
<div class="gmail_extra"><br>
<div class="gmail_quote">On Sep 18, 2017 7:51 PM, William Schuler <schulerw@gmail.com> wrote:<br type="attribution">
<blockquote class="quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div><font size="2"><span style="font-size:10pt">
<div>hi all,<br>
<br>
i don't see us discussing four papers in one day. evan, can we perhaps triage a couple of these papers (say #3 and #4) and discuss when we meet whether to reorganize our schedule to cover them on a subsequent day?<br>
<br>
wm<br>
<br>
> On Sep 18, 2017, at 3:13 PM, Evan Jaffe <jaffe.59@buckeyemail.osu.edu> wrote:<br>
> <br>
> Hey all, remember we're doing 4 memory papers this week? Yeah, I forgot<br>
> too - start reading now ;)<br>
> <br>
> Sukhbaatar et al 2015: End-to-end Memory Networks<br>
> <a href="https://arxiv.org/abs/1503.08895">https://arxiv.org/abs/1503.08895</a><br>
> <br>
> Kaiser et al 2017: Learning to Remember Rare Events<br>
> <a href="https://arxiv.org/abs/1703.03129">https://arxiv.org/abs/1703.03129</a><br>
> <br>
> Kumar et al 2016: Ask Me Anything: Dynamic Memory Networks for NLP<br>
> <a href="https://arxiv.org/abs/1506.07285">https://arxiv.org/abs/1506.07285</a><br>
> <br>
> Miller et al 2016: Key-Value Memory Networks For Directly Reading<br>
> Documents <a href="https://arxiv.org/abs/1606.03126">https://arxiv.org/abs/1606.03126</a><br>
> <br>
> <br>
> I swapped the original Weston out for the Sukhbaatar, which seems a<br>
> little more polished and readable. Everyone cites it too, so it's #1<br>
> priority for discussion. (feel free to look at Weston 2015, Memory<br>
> Networks as well)<br>
> <br>
> The Kaiser is the most recent and has some nice decay-like properties,<br>
> and that'll be priority 2.<br>
> <br>
> Kumar is 3rd and extends earlier Weston models to sequences,<br>
> incorporating positional information. Haven't looked at it in detail yet.<br>
> <br>
> The Miller adds some encoding hashing something to the Sukhbaatar, so it<br>
> seems like an incremental improvement given my short skim - lowest<br>
> priority unless I get lots of angry mail with convincing arguments to<br>
> the contrary.<br>
> <br>
> See you guys Thursday,<br>
> <br>
> Evan<br>
> <br>
> <br>
> <br>
> <br>
> <br>
> <br>
> <br>
> <br>
> _______________________________________________<br>
> CaCL mailing list<br>
> CaCL@lists.osu.edu<br>
> <a href="https://lists.osu.edu/mailman/listinfo/cacl">https://lists.osu.edu/mailman/listinfo/cacl</a><br>
</div>
</span></font></div>
</blockquote>
</div>
<br>
</div>
</body>
</html>