internlp2021/index.html at main · Wuhn/internlp2021 · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
<!DOCTYPE html>
<html>
<head>
	<style>
	  th {font-family:Verdana;}
	  p {font-family:Verdana;}
	  tr.grey {background-color: #ddd;}
	  tr.darkgrey {background-color: #eee;}
	</style>
</head>
<body onload="updateTimeZone()">

<script>
var daylightSavingsAdjust = 0;

function updateTimeZone(){
  var lstCity = document.timezone.lstCity;
  var timeDiff = lstCity.options[lstCity.selectedIndex].value;
  updateTime(timeDiff);
}

function getHours(timeString){
  var hours = timeString.split(":")[0];
  return hours;
}

function getMinutes(timeString){
  var minutes = timeString.split(":")[1];
  return minutes;
}

function addTime(defaultTime, difference){
	var minutes = getMinutes(defaultTime);
	var hours = Number(getHours(defaultTime)) + Number(difference);

	var finalTime = ""
	if (hours > 24) {
		var diff = hours-24;
		if (diff < 10){
			finalTime = "0" + diff.toString() + ":" + minutes + " (6th of August) ";
		} else {
			finalTime = diff.toString() + ":" + minutes + " (6th of August) ";
	}	} else if (hours < 0) {
		var diff = 24+hours;
		if (diff < 10){
			finalTime = "0" + diff.toString() + ":" + minutes ;
		} else {
			finalTime = diff.toString() + ":" + minutes ;
		}
	} else {

		if (hours < 10) {
			finalTime = "0" + hours.toString() + ":" + minutes;
		} else {
			finalTime = hours.toString() + ":" + minutes;
		}
	}

	return finalTime;
}

function updateTime(timeDiff){
  let utc0 = document.getElementById("session").querySelectorAll("span");
  let utcx = document.getElementById("session").querySelectorAll("p");

for (let i = 0; i < utc0.length; i++) {
	let defaulttime = utc0[i];
	let targettime = utcx[i];

	var resultsText = addTime(defaulttime.innerHTML, timeDiff);
	targettime.innerHTML = resultsText;
  };
}

function updateTable1(timeDiff){

  let utc0 = document.getElementById("session1").querySelectorAll("span");
  let utcx = document.getElementById("session1").querySelectorAll("p");

for (let i = 0; i < utc0.length; i++) {
	let defaulttime = utc0[i];
	let targettime = utcx[i];

	var resultsText = addTime(defaulttime.innerHTML, timeDiff);
	targettime.innerHTML = resultsText;
  };
}

function updateTable2(timeDiff){

  let utc0 = document.getElementById("session2").querySelectorAll("span");
  let utcx = document.getElementById("session2").querySelectorAll("p");

for (let i = 0; i < utc0.length; i++) {
	let defaulttime = utc0[i];
	let targettime = utcx[i];

	var resultsText = addTime(defaulttime.innerHTML, timeDiff);
	targettime.innerHTML = resultsText;
  };
}

function updateTable3(timeDiff){

  let utc0 = document.getElementById("session3").querySelectorAll("span");
  let utcx = document.getElementById("session3").querySelectorAll("p");

for (let i = 0; i < utc0.length; i++) {
	let defaulttime = utc0[i];
	let targettime = utcx[i];

	var resultsText = addTime(defaulttime.innerHTML, timeDiff);
	targettime.innerHTML = resultsText;
  };
}

</script>

<table cellspacing="2" cellpadding="4" style="font-family:Verdana" id="session1">
  <tr>
	<th>Timezones</th>
	<td>
	<form name="timezone">
	  <select name="lstCity" onchange="updateTimeZone();">
	  <option value="-12">UTC -12
	  <option value="-11">UTC -11
	  <option value="-10">UTC -10
	  <option value="-9">UTC -9
	  <option value="-8">UTC -8
	  <option value="-7">UTC -7
	  <option value="-6">UTC -6
	  <option value="-5">UTC -5
	  <option value="-4">UTC -4
	  <option value="-3">UTC -3
	  <option value="-2">UTC -2
	  <option value="-1">UTC -1
	  <option value="0" selected>UTC +0
	  <option value="1">UTC +1
	  <option value="2">UTC +2
	  <option value="3">UTC +3
	  <option value="4">UTC +4
	  <option value="5">UTC +5
	  <option value="6">UTC +6
	  <option value="7">UTC +7
	  <option value="8">UTC +8
	  <option value="9">UTC +9
	  <option value="10">UTC +10
	  <option value="11">UTC +11
	  <option value="12">UTC +12
	  </select>
	</form>
	</td>
  </tr>
</table>

<table cellspacing="2" cellpadding="4" style="font-family:Verdana" id="session">
  <tr>
    <th style="min-width:80px">Session 1</th>
  </tr>
  <tr class="grey">
    <td><p id="session1-1">12:00</p><span id="session1-1" hidden>12:00</span></td>
    <td tyle="min-width:100px">Opening Remarks</td>
	<td></td>
  </tr>
  <tr class="darkgrey">
    <td ><p id="session1-2">12:05</p><span id="session1-2" hidden>12:05</span></td>
	<td>Invited talk</td>
    <td>
    <details>
	<summary>Dan Goldwasser: <a href="https://youtu.be/eG2umIzyP0Q" target="_blank" rel="noopener noreferrer">Using Neuro-Symbolic Approaches to Bridge the Gap between Human and Machine Representations</a></summary>
	The interaction between humans and machines requires a set of symbols understood in a similar way by both sides. As neural language representations become increasingly complex, ensuring that common understanding becomes significantly harder. In this talk, I’ll discuss DRaiL, a declarative neuro-symbolic framework designed to help support such interactions by taking a scaffolded language learning approach, using intermediate relational concepts. The talk will focus on several recent applications for political discourse analysis.</details>
    </td>
  </tr class="grey">
  <tr>
    <td valign="top"><p id="session1-3">12:50</p><span id="session1-3" hidden>12:50</span></td>
    <td>Poster short talks</td>
	<td><a href="https://aclanthology.org/2021.internlp-1.4/">A Proposal: Interactively Learning to Summarise Timelines by Reinforcement Learning</a>, Yuxuan Ye and Edwin Simpson.</br>
		<a href="https://aclanthology.org/2021.internlp-1.6/">Active Curriculum Learning</a>, Borna Jafarpour, Dawn Sepehr and Nick Pogrebnyakov
	</td>
  </tr>
  <tr class="darkgrey">
    <td><p id="session1-4">13:20</p><span id="session1-4" hidden>13:20</span></td>
    <td>Poster session</td>
	<td>The poster session will happen in gather.town setup by Underline. <br>
	    Please click on the gather.town button on the <a href="https://underline.io/events/167/sessions?eventSessionId=6290">underline workshop page</a> and navigate to the workshop room.
	</td>
  </tr>
  <tr class="grey">
    <td class="second" valign="top"><p id="session1-5">14:30</p><span id="session1-5" hidden>14:30</span></td>
	<td>Invited talk</td>
    <td><details>
	  <summary>Dorsa Sadigh: <a href="https://youtu.be/20qw6rCVF30" target="_blank" rel="noopener noreferrer">Beyond Imitation Learning: New  Paradigms of Querying and Integrating Different Types of Human Data for Improved Robot Learning</a></summary>
	  Learning from humans is a fundamental component of building better robotic algorithms that can achieve complex tasks. This plays a central role in specifying how dynamical systems should act, i.e., how an assistive robot arm should move, or how an autonomous car should drive. Learning such models is highly dependent on access to a large amount of expert data collected from humans interacting with robots. However, we usually do not have access to such a large dataset in robotics settings, and even the data provided can be noisy, inaccurate, repetitive, or not informative. In this talk, we will discuss two core ideas to effectively learn from interactions with humans. First, we discuss how to develop active learning techniques that query humans in a directed manner in order to learn from them. Second, we discuss how we can tap into many different sources of data beyond expert demonstrations to learn in an interactive fashion from humans. We discuss some of our recent work on learning from physical corrections, preferences, and suboptimal human data. We end the talk by discussing a new thread of work on integrating language in the robot learning loop to enable language-enabled shared autonomy systems for assistive teleoperation.
    </details>
	</td>
  </tr>
  <tr class="darkgrey">
    <td><p id="session1-6">15:15</p><span id="session1-6" hidden>15:15</span></td>
    <td>End</td>
	<td></td>
  </tr>
  <tr>
    <th style="min-width:80px">Session 2</th>
  </tr>
  <tr class="grey">
    <td><p id="session2-1">16:00</p><span id="session2-1" hidden>16:00</span></td>
    <td>Opening Remarks</td>
	<td></td>
  </tr>
  <tr class="darkgrey">
    <td valign="top"><p id="session2-2">16:05</p><span id="session2-2" hidden>16:05</span></td>
	<td>Invited talk</td>
    <td><details>
		<summary>Yoav Artzi: <a href="https://youtu.be/_XiYWKAo7Ig" target="_blank" rel="noopener noreferrer">Continual Language Learning through Collaborative Interaction with Users</a></summary>
		Interaction with human users provides ample opportunities for language learning. I will describe how collaborative scenarios provide a strong foundation for learning from interaction with users, by combining strong incentives for teaching and flexibility for adaptation. I will show how this applies to instruction generation learning, where a system learns by comparing human execution of generated instructions and their original system intents to infer the system's communicative success. We use this signal to formulate a contextual bandit learning problem, where the system continually improves over time through interaction with users. In interaction with real users, we demonstrate dramatic improvements in the system's ability to generate language over time.
    </details>
	</td>  </tr>
  <tr class="grey">
    <td><p id="session2-3">16:50</p><span id="session2-3" hidden>16:50</span></td>
    <td>break</td>
	<td></td>
  </tr>
  <tr class="darkgrey">
    <td><p id="session2-4">17:00</p><span id="session2-4" hidden>17:00</span></td>
    <td>Panel discussion</td>
	<td>Julia Kreutzer, Alison Smith-Renner, Ido Dagan <br>
	<b>Considering the user in design and evaluation of interactive NLP systems.</b></td>
  </tr>
  <tr class="grey">
    <td><p id="session2-5">17:45</p><span id="session2-5" hidden>17:45</span></td>
    <td>break</td>
	<td></td>
  </tr>
  <tr class="darkgrey">
    <td><p id="session2-6">18:00</p><span id="session2-6" hidden>18:00</span></td>
    <td>Poster short talks</td>
	<td>
	<a href="https://aclanthology.org/2021.findings-acl.54/">Dialogue in the Wild: Learning from a Deployed Role-Playing Game with Humans and Bots</a>, Kurt Shuster, Jack Urbanek, Emily Dinan, Arthur Szlam and Jason Weston. </br>
	<a href="http://proceedings.mlr.press/v139/nguyen21e.html">Interactive learning from activity description</a>, Khanh Nguyen, Dipendra Misra, Robert Schapire, Miro Dudík and Patrick Shafto. </br>
	<a href="https://aclanthology.org/2021.internlp-1.7/">Tackling Fake News Detection by Interactively Learning Representations using Graph Neural Networks</a>, Nikhil Mehta and Dan Goldwasser.
	</td>
  </tr>
  <tr class="grey">
    <td><p id="session2-7">18:30</p><span id="session2-7" hidden>18:30</span></td>
    <td>Poster session</td>
	<td>The poster session will happen in gather.town setup by Underline. <br>
	    Please click on the gather.town button on the <a href="https://underline.io/events/167/sessions?eventSessionId=6290">underline workshop page</a> and navigate to the workshop room.
	</td>
  </tr>
  <tr class="darkgrey">
    <td><p id="session2-8">19:45</p><span id="session2-8" hidden>19:45</span></td>
    <td>break</td>
	<td></td>
  </tr>
  <tr class="darkgrey">
    <td><p id="session2-9">20:00</p><span id="session2-9" hidden>20:00</span></td>
	<td>Invited talk</td>
    <td>Percy Liang: <a href="https://youtu.be/AYeRirULqgk" target="_blank" rel="noopener noreferrer">Interacting with Deep Learning Models</a></td>
	<td></td>
  </tr>
  <tr  class="grey">
    <td><p id="session2-10">20:45</p><span id="session3-5" hidden>20:45</span></td>
    <td>End</td>
	<td></td>
  </tr>
  <tr>
    <th style="min-width:80px">Session 3</th>
  </tr>
  <tr class="grey">
    <td class="first"><p id="session3-1">23:00</p><span id="session3-1" hidden>23:00</span></td>
    <td class="first">Opening Remarks</td>
	<td></td>
  </tr>
  <tr class="darkgrey">
    <td><p id="session3-2">23:05</p><span id="session3-2" hidden>23:05</span></td>
	<td>Invited talk</td>
    <td>
		Dan Roth: <a href="https://youtu.be/JXSWJVENM0c" target="_blank" rel="noopener noreferrer">Reasoning Interactively about Text</a>
	</td>
  </tr>
  <tr class="grey">
    <td><p id="session3-3">23:50</p><span id="session3-3" hidden>23:50</span></td>
    <td>break</td>
	<td></td>
  </tr>
  <tr  class="darkgrey">
    <td><p id="session3-4">24:00</p><span id="session3-4" hidden>00:00</span></td>
    <td>Panel discussion</td>
	<td>Seung-Won Hwang, Dan Roth, Sherry Tongshuan Wu
	<br> <b>Machines in the loop: explainability, transparency, and rich interaction.</b></td>
  </tr>
  <tr  class="grey">
    <td><p id="session3-5">24:45</p><span id="session3-5" hidden>00:45</span></td>
    <td>End</td>
	<td></td>
  </tr>
</table>

</body>
</html>